CGM for healthy people: Useful N=1 toy, or marketing in a sticker?

A CGM measures glucose in the fluid between your cells, not in your blood, and estimates your blood value from there. The sensor is a thin filament under the skin of your upper arm, so what it reads is the interstitial fluid (the watery space around your cells). That fluid trails your blood by a few minutes, about 6 minutes in one independent test of over-the-counter sensors.^[17]

The gap between sensor and blood grows when your sugar rises or falls fast, and in that test readings were least reliable right after a sugary drink.^[17] So after a meal or a hard sprint, your curve and a finger-prick taken in the same second can disagree. Lag explains many of these gaps, but not all of them, because sensors can also simply be off and a recalled batch can read falsely low.^[23] If a reading surprises you or does not match how you feel, check it with a finger-prick.

That recall covers some FreeStyle Libre 3 and Libre 3 Plus sensors that can read falsely low.^[23] It began in November 2025, when Abbott wrote to affected customers, and the FDA issued an early alert in December 2025.^[23] In February 2026 the FDA classified it as a Class I recall, its most serious category.^[23] Check the serial number of every sensor you wear or have at home at freestylecheck.com, stop using and discard any affected sensor, and request the free replacement there.^[23] The danger is greatest for people who dose insulin or carbohydrates by the reading, but a false low can also send a healthy person off to eat food they do not need.

Even a sensor from a good batch misses by a few percent. Researchers express that as MARD, the average percentage by which readings miss a reference value, so lower is better. In a head-to-head test in adults with diabetes, paid for by Libre maker Abbott, the FreeStyle Libre 3 scored 8.9 percent and the Dexcom G7 13.6 percent.^[7] MARD is an average rather than a limit, so the more useful figure is how many readings land close. About 9 in 10 Libre 3 readings and 8 in 10 G7 readings were within 20 mg/dL (or 20 percent) of the lab value.^[7]

People without diabetes were not in that test. The first independent test of over-the-counter sensors in this group, in 39 young adults, found that readings missed a finger-prick value by about 13 percent on average.^[17] They drifted most when glucose moved fast.^[17] The authors advise using the sensors for trends and being cautious with decisions at thresholds, such as spotting a low or screening yourself for diabetes.^[17] A CGM is good at showing the shape of your day and weaker at settling any single value, and that shape is what the studies of healthy people describe.

Without diabetes, a CGM average of about 99 mg/dL (milligrams of glucose per decilitre) is normal for non-obese, mostly younger people.^[3] For middle-aged adults with normal glucose, about 115 mg/dL is normal.^[10] The 99 comes from a study at 12 US centres that put blinded sensors (screen hidden) on 153 healthy, non-obese people aged 7 to 80, more than a third of them children and teens.^[3] Most of their day stayed between 70 and 140 mg/dL and almost none of it above 180.^[3] Even the young ones crossed 140 now and then, mostly during the day and hardly ever at night.^[3]

The 115 comes from a larger community study that measured adults without diabetes once, at an average age of about 58.^[10] Among those with normal glucose, most had 24-hour averages between 100 and 140 mg/dL, and they spent roughly 3 hours a day above 140.^[10] Time above 180 was rare but not zero, since about one in five of them spent more than 2 percent of the day there.^[10] A single excursion above 180 is not a finding, but long stretches there are worth a doctor's look.

The authors of the larger study offer a reason for the gap between the two averages. The earlier normal values, they note, came from smaller groups of non-obese people whose sensors were calibrated by finger-prick every day.^[10] So a 55-year-old who sees 115 on the screen is in line with this group, not running high. Brief rises above 140 mg/dL during the day are normal in both studies, and how often they happen depends on age and weight.^{[3, 10]}

The table puts the two groups side by side, first the non-obese, mostly young people from the 12-centre study, then the adults around 58.

Sources: Shah et al. 2019^[3]; Spartano et al. 2025^[10]

These glucose averages describe what is common, not an official target for healthy people, because no such target exists. The international time-in-range targets were written for people with diabetes.^[1] Time in range itself, the share of the day spent inside a set band, was developed to complement HbA1c (a three-month blood-sugar average) in diabetes care.^[8] The ADA Standards of Care 2025 describe over-the-counter CGMs as meant for people with prediabetes or diabetes not on insulin and make no recommendation for healthy people.^[4] The 2026 update adds that anyone, including people without diabetes, can buy them to see how food and exercise move their glucose, still without a clinical recommendation.^[5] That makes the table a yardstick for reading your own two weeks rather than a score to beat.

In two weeks a CGM can show you how your own curve responds to specific meals, to exercise, alcohol, stress or a cold, and what it looks like after a short night. You can, for example, compare white bread on its own with the same bread after a walk. On another day, the same bread alongside some protein and fat gives you a second comparison. Whatever difference shows up is a personal observation, your own data rather than a population average.

Sleep is the hardest of these to read. In a study of 953 healthy adults given standardised breakfasts, poorer sleep efficiency and later bedtimes were linked to larger blood-sugar rises after breakfast.^[11] Sleep efficiency is the share of time in bed actually spent asleep. In your own curve that effect is hard to separate from what you ate and did the evening before, so a single bad morning tells you little.

What a CGM cannot do is diagnose, because the consumer sensors are explicitly not meant for it.^[27] Prediabetes and diabetes are diagnosed with fasting plasma glucose, HbA1c or an OGTT, a sugar-drink test with a blood draw two hours later.^[19] Nor can it tell you whether a spike to 150 mg/dL harms you. No randomized trial shows that acting on CGM data improves heart or longevity outcomes in healthy people.^{[15, 16]} And it cannot tell you to go low-carb, keto or start intermittent fasting; those are decisions for a talk with a Hausarzt (your GP) or a registered dietitian.

Used as an experiment on one person, you, a CGM can help you spot patterns worth bringing to a doctor. Treated as a daily report card, it may feed anxiety and needless food restriction. For a few groups of people that risk, or a missed low, outweighs the curiosity from the start.

A CGM bought out of curiosity is not for people with a current or past eating disorder or for anyone who takes insulin or other drugs that can cause lows. Nor is it for anyone hoping it will rule out diabetes. These groups need a doctor in the loop before the sensor goes on:

• If you have or had an eating disorder, talk to a doctor first; Abbott's own terms for its Lingo sensor say the same.^[27]
• If you take insulin or other glucose-lowering drugs that can cause lows (for example sulfonylureas), use a prescribed system with alarms and your diabetes team, because wellness sensors such as Lingo and Stelo give no low-glucose alerts.^{[5, 20]}
• Stelo is not meant for people with problematic hypoglycaemia (frequent or severe lows).^[20]
• A CGM cannot rule out diabetes, because the consumer devices are explicitly not for diagnosis.^[27] In an independent test their single readings agreed only loosely with finger-prick values, so the authors advise caution with threshold decisions such as spotting a low or screening for diabetes.^[17]
• Symptoms such as strong thirst, frequent urination or unexplained weight loss need a doctor and a blood test, not a sensor.

A low reading is where the finger-prick matters most, because a low CGM reading alone does not prove a low.^{[17, 23]} Check an unexpected one with a finger-prick, even if you feel fine. If the finger-prick confirms a value below 70 mg/dL and you feel shaky, sweaty or dizzy, follow these steps.^[30]:

1. First make sure you are awake and can swallow safely. Anyone who is drowsy, very confused, cannot swallow or passes out gets nothing by mouth and needs emergency help (112).
2. Eat or drink about 15 g of fast carbohydrate, such as juice or glucose tablets^[30]
3. Recheck after 15 minutes.^[30]
4. If the value is still below 70 mg/dL, take another 15 g.^[30]

Repeated confirmed lows belong with your Hausarzt. Everyone else can try a short experiment with the recall and finger-prick checks in mind, as long as they expect data about their meals rather than the health gains the marketing promises.

The marketing gets ahead of the science where it promises better health, because no randomized trial shows that wearing a CGM improves heart or longevity outcomes in healthy people.^{[15, 16]} What it builds on is one real finding, that blood-sugar responses to the same meal vary from person to person.^[2] Lingo, Stelo and coaching apps such as Levels, NutriSense and Hello Inside all start from it. The step from that finding to claims about metabolic health and longevity is the part the evidence does not yet carry.

That finding goes back to a Stanford study that coined the word "glucotypes".^[2] The team ran CGMs on 57 adults and gave standardised test meals to a subset of them. Responses to the same meal varied widely, and three patterns of low, moderate and severe swings showed up, including in people whose HbA1c (a three-month blood-sugar average) was normal.^[2]

Those 57 people were not all healthy, though. Among those given the test meals, some turned out to have prediabetes or diabetes, and the three glucotype groups also differed in average glucose.^[2] A later critique built on that point and questioned whether glucotypes show more than average glucose and whether they hold up in other groups.^[12] The variation between people is real, but the study was observational (it watched people rather than testing a change) and tracked no heart or longevity outcome.

The regulatory green light is often read as the missing proof, but it is a permission to sell. In 2024 the FDA cleared two of the first over-the-counter CGMs, Dexcom Stelo and Abbott Lingo.^{[20, 22]} A clearance means a device is safe enough to sell off the shelf, not that wearing one makes a healthy person healthier.

The reviews of people without diabetes find little that would close that gap. A 2024 review found no consistent, high-quality evidence of benefit and said commercial claims for people without diabetes should be labelled misleading.^[6] Its authors also called for stronger post-market oversight of the devices.^[6] A 2026 meta-analysis, which pools the results of many studies, narrows the picture further.^[15] It suggests that CGM feedback helps mostly people with prediabetes, with no clear glucose benefit in people whose glucose is already normal.^[15] Most trials in both 2026 reviews ran for weeks or months, and none measured hard outcomes such as heart attacks, new diabetes or death.^{[15, 16]} The longest we found followed people with prediabetes for two years and measured HbA1c.^[37]

Weight is the one area with a little trial data, and it comes from people with overweight or obesity. Two small, short randomized trials found extra weight loss with CGM feedback, one alongside dietary education and one within a diet and exercise programme.^{[13, 14]} Even so, the 2026 meta-analysis found no clear effect on BMI and concluded that a CGM is unlikely to help with weight on its own, without a behaviour programme around it.^[15]

The other pillar of the marketing, that glucose swings harm the heart, comes from a different population. A 2021 review linked glucose swings to heart outcomes in people with type 2 diabetes.^[9] It says nothing about healthy people, so carrying the finding over to them is an inference.

The downsides are possible rather than proven, which is not the same as harmless. Watching every spike can feed food anxiety or needless restriction, the slide toward orthorexia (an obsession with eating "correctly"). How often that happens is unknown. The 2024 review flagged effects on eating habits as poorly studied,^[6] and a 2026 review found the psychological evidence too thin to show either benefit or harm.^[16] Anchoring is the everyday version, as when someone sees 145 mg/dL after porridge, switches to bacon and eggs and never questions the swap again.

With the health case still open, what a healthy person buys is feedback on their meals. Whether that feedback is worth paying for depends on the sensor price and how much guidance you want.

In Germany a healthy adult has three self-paid routes: FreeStyle Libre 3 Plus sensors without a prescription, the pricier Dexcom G7, or Hello Inside, which sells Libre sensors with a coaching app. The table shows the prices as of October 2026.^{[24, 25, 26]} Per day of wear, the Libre 3 Plus bought from Abbott costs least, while the coaching package costs extra. Prices move, so check them on the day you buy; the US consumer sensors are listed for comparison.

Sources: manufacturer and provider pages [24–28], accessed October 2026.

All of these routes are self-paid because of the way cover works. Statutory insurance (GKV) in Germany pays for a CGM such as the Libre 3 Plus only for type 1 or type 2 diabetes treated with intensified insulin therapy^[32] Even then, each insurer decides case by case.^[32] The basis is a 2016 decision of the G-BA, the body that sets statutory cover, on real-time CGM for insulin-requiring diabetes.^[36] Private insurers decide by contract, and without such cover you pay yourself (as of October 2026).

Digital health apps (DiGA) do not change that. Glucura, for example, is meant for diagnosed type 2 diabetes in people who do not use insulin.^[31] Insurers pay for a DiGA on prescription or, with proof of the diagnosis, after approving an application,^[38] so neither route is open to healthy users. This guide checked cover and prices for Germany, so in Austria and Switzerland ask your insurer before assuming anything.

Dexcom G7 is a diabetes device, but in Germany self-payers can also buy it without a prescription in Dexcom's own shop.^[25] Per day that comes to about EUR 9.90, roughly twice the Libre 3 Plus, so it makes sense mainly if you already know and prefer the Dexcom system.^{[24, 25]}

Hello Inside puts its coaching app on top of a FreeStyle Libre sensor whose model depends on the country.^[26] Besides the sensor and the two months of app access in the table, its starter package includes trial access to its premium coaching.^[26] Each further sensor works out at about EUR 6.40 a day against about EUR 4.70 for Abbott's own sensor (as of October 2026).^{[24, 26]} Whether the app is worth that extra depends on how much guidance you want in reading your curve. Levels and NutriSense in the US follow a similar model of a third-party sensor plus coaching. Levels runs from about USD 15 a month for the app alone to about USD 111 a month for its top plan.^[29]

Lingo and Stelo, the two US consumer sensors, are not officially sold here. Lingo builds on Abbott's Libre sensor technology and is sold in the US and the UK.^[22] As of October 2026 Abbott has not launched it in Germany, Austria, Switzerland or elsewhere in the EU.^[27] People here get it only through cross-border parcels or reshippers, which is a generous definition of availability.^[27] Stelo is Dexcom's equivalent, sold in the US as a USD 89 monthly subscription with 15-day sensors.^[28] In the US, both are meant only for people who do not use insulin:

• Dexcom Stelo: cleared by the FDA in March 2024 for adults not on insulin,^[20] and in June 2026 also for children from age 2 who do not use insulin.^[21]
• Abbott Lingo: FDA clearance announced by Abbott in June 2024.^[22]; sold over the counter to adults 18 and up who do not use insulin^[27]

For a healthy adult in Germany, the realistic path is one Libre 3 Plus sensor for a two-week test, or two for a month at about EUR 140.^[24] A coaching service is worth adding only if you want the guidance. Set against what a year of sensors costs, the blood tests that diagnose problems look like a bargain.

The tests that actually diagnose a blood-sugar problem are fasting plasma glucose, HbA1c and, when indicated, a 75 g OGTT.^[19] In that test, glucose is measured fasting and again two hours after a sugar drink.^[19] HbA1c reflects your average blood sugar over roughly the past two to three months, while fasting glucose shows your level on the morning of the test.^[19] These tests are boring and cheap, which in lab medicine counts as a compliment. They are what a doctor uses to diagnose prediabetes and diabetes, something no consumer CGM is built for.^{[19, 27]}

For heart risk, other markers have decades of outcome research behind them. ApoB counts the particles that carry cholesterol into the artery wall, and studies show that fewer of these particles mean fewer heart attacks and strokes.^[33] Lp(a) is a largely inherited blood-fat marker, and large genetic and observational studies support a causal role in heart and blood-vessel disease.^[35] European experts advise measuring it at least once in adult life.^[35] Blood pressure belongs on the list too, because lowering it prevents cardiovascular events even in people whose values are not markedly high.^[34]

Fasting insulin and the HOMA-IR score calculated from it are cheap estimates of insulin resistance, which means your cells respond less well to insulin. Some doctors add them, but a 2023 laboratory guideline advises against routine insulin testing in people at risk of diabetes or heart disease.^[18] Its reasons are that insulin values add little to risk prediction and have no agreed cut-offs.^[18]

A few extras sit outside routine care. The exact measurement of insulin resistance, the clamp test, is a research method.^[18] An OGTT with extra insulin samples is an add-on some specialists use, not the standard, and hs-CRP (a sensitive inflammation marker) belongs in the same optional group.

A once-a-year private panel with HbA1c, fasting glucose and lipids including ApoB costs roughly EUR 150 to 350, depending on lab and markers. A year of Libre 3 Plus sensors alone comes to about EUR 1,700.^[24] Before you pay out of pocket for either, check what your health insurance already covers. If you are curious about your curve anyway, our reading of these studies favours one planned two-week experiment on your meals, movement and sleep, with blood tests and a doctor for any health question.