A continuous glucose monitor is a small filament sensor sitting in the subcutaneous fat of your upper arm. The first surprise: it does not measure blood glucose. It measures glucose in the interstitial fluid — the watery space between cells — and estimates your blood value from there. Interstitial glucose lags capillary blood by roughly 5 to 15 minutes, longer during sharp rises or drops, which is why a CGM curve can look noticeably different from a finger-prick reading at the same second.

Accuracy is reported as MARD — Mean Absolute Relative Difference. A 2024 head-to-head by Hanson and colleagues in 55 insulin-treated adults with type 1 or type 2 diabetes reported point-accuracy MARD of 8.9 percent for FreeStyle Libre 3 and 13.6 percent for Dexcom G7 against laboratory reference; accuracy in healthy normoglycemic ranges may differ. Manufacturer pivotal trials report different numbers (often closer to 8 to 9 percent for both), because population, reference method and glucose range matter enormously. Practical takeaway: a value of 95 mg/dL could plausibly be anywhere from 86 to 104 mg/dL in real blood. Two sensors on the same arm will disagree, sometimes by 15 mg/dL. That is the technology working as designed.

Devices in DACH: Abbott FreeStyle Libre 3, Dexcom G7 and Dexcom ONE+ (CE-marked, primarily prescribed for diabetes); Abbott Lingo (consumer biowearable, UK and US launch 2024); Dexcom Stelo (US 2024, European rollout ongoing). Platforms like Hello Inside (Austrian start-up, Vienna) pair a third-party sensor with a coaching app.

The most important single dataset is Shah and colleagues 2019, published in the Journal of Clinical Endocrinology and Metabolism. They put CGMs on 153 nondiabetic adults across multiple US sites and recorded continuous glucose for several days. The mean 24-hour glucose was 99 ± 7 mg/dL. The median percentage of time spent between 70 and 140 mg/dL was 96 percent (IQR 93–98), and median time above 180 mg/dL was effectively zero. People aged 60 and over spent measurably more time above 140 mg/dL than younger adults, but even healthy young adults occasionally crossed 140 — often briefly after meals.

The Spartano 2025 community cohort published in the same journal extended this with a much larger group of adults without diabetes and confirmed the same picture: in healthy people, time-above-180 mg/dL is rare, time-in-range (70 to 180 mg/dL) typically exceeds 95 to 99 percent, and the overall distribution is narrower than commercial CGM apps tend to suggest.

The most-cited paper for consumers is Hall and colleagues 2018, published in PLOS Biology, which coined the term glucotypes. The Stanford team used CGM and standardized meal tests on 57 adults and found that postprandial responses to identical meals varied widely between people — three distinct response patterns (low, moderate, severe variability) emerged even in adults classified as normoglycemic by HbA1c. This is the source of the now-popular claim that everyone responds differently to food. It is true, and it is interesting. It is also a 57-person observational study without longevity or cardiovascular outcomes attached.

The ATTD international consensus (Battelino 2019, Diabetes Care) defines clinical CGM targets but explicitly for people with diabetes. There is no consensus on what time-in-range means or matters for non-diabetic adults. The ADA Standards of Care 2025 (Section 7) limits its OTC-CGM recommendation to people with prediabetes or with diabetes not on insulin; it does not endorse CGM as a routine longevity or wellness tool in healthy adults. The 2026 update has since acknowledged that OTC biosensors are also purchased by people without diabetes who want to observe glycemic responses to lifestyle — still without a clinical recommendation for healthy users.

Honest list of things a CGM can reveal in a healthy person, with caveats. First, meal-by-meal differences in your own postprandial curves: white bread on its own versus the same bread eaten after a walk or alongside protein and fat will produce different rises. This is your data, not a population average. Second, the effect of poor sleep on next-morning fasting glucose and tolerance — highlighted in the Oganesova 2024 review as a reproducible insight from consumer CGM use. Third, what alcohol, illness, stress and intense exercise do to your curve.

There is a long list of what a CGM cannot tell you. It cannot diagnose prediabetes or diabetes — those need fasting glucose, OGTT and HbA1c, not CGM peaks. It cannot tell you that a spike to 150 mg/dL is bad. It cannot tell you whether your spike pattern predicts heart disease or mortality 20 years from now, because no randomized trial in healthy adults has tested that. And it cannot tell you to go low-carb, keto, carnivore or fasting — those are clinical decisions that need a Hausarzt or registered dietitian.

The most honest framing: a CGM is a structured n=1 experiment. If you treat it as a curiosity tool to notice patterns and discuss anomalies with a physician, it can be useful. As a daily report card, it more often produces anxiety, food avoidance and overcorrection that is itself unhealthy.

The CGM-for-wellness market is a textbook example of a strong biomarker chasing a much weaker outcome story. Abbott (Lingo), Dexcom (Stelo), Levels Health (US-only, not available in DACH), Hello Inside (Austria), Veri (Finland), Glucura and Una Health (Germany) all build on a real signal — postprandial glucose varies between people and foods — and extrapolate to claims about metabolic health and longevity that the evidence does not yet support.

The 2024 narrative review by Oganesova, Pemberton and Brown in Diabetic Medicine looked specifically at non-diabetic CGM use and reached a notably cautious verdict: evidence for utility is limited and inconsistent, commercial claims risk being misleading, potential unintended adverse effects (over-restriction, anxiety) warrant attention, and the authors call for stricter regulation of consumer marketing. Martinez et al. 2021 in BMJ Open Diabetes Research and Care is similarly careful: glucose variability associates with cardiovascular outcomes in type 2 diabetes, but extrapolating to healthy adults is not warranted.

The risks are not zero. The anchoring effect is real: people see a 145 mg/dL spike after porridge, panic, switch to bacon and eggs, never re-examine the assumption. The orthorexia risk is real: case reports link CGM use to disordered eating in already-anxious users. Cost side: a healthy DACH self-pay user spends EUR 60 to 120 per month, EUR 700 to 1500 per year — none reimbursed by GKV or private insurance without a coded diabetes diagnosis under ICD-10 E10 or E11.

Honest summary: CGM in healthy people is an interesting biomarker, not a validated longevity intervention. Treat it like a sauna habit you are testing — try it, see if it changes anything you actually care about, then decide whether EUR 1000 per year buys more than a single private blood panel with HbA1c, fasting insulin and OGTT.

The DACH CGM landscape in mid-2026 is a moving target; check current availability at purchase time. Here is the neutral overview.

Abbott FreeStyle Libre 3 is the dominant prescription CGM in Germany and Austria. It is on the Hilfsmittelverzeichnis (HMV) and reimbursed by statutory health insurance only for people with a coded diabetes diagnosis under intensive insulin therapy or comparable indication. Without a diabetes diagnosis, you can buy sensors privately through an Apotheke (Selbstzahler) at roughly EUR 60 to 75 per 14-day sensor. In Switzerland, CGM is covered through the Grundversicherung under KLV/MiGeL only for diagnosed diabetes meeting specific clinical criteria; non-diabetic wellness use is not covered.

Abbott Lingo is the consumer biowearable, launched OTC in the United Kingdom and United States in 2024. As of 2026 its DACH rollout is partial and channel-dependent — some users ship from UK, others wait for local availability. CE-MDR status for direct consumer sale in DE/AT/CH should be verified at point of purchase.

Dexcom G7 is prescription-only in DACH for diabetes management. Dexcom Stelo, the US OTC product launched 2024, has not yet seen a full DACH consumer launch as of mid-2026; check the manufacturer's regional pages.

Hello Inside is a Vienna-based start-up offering a CGM-plus-coaching subscription marketed primarily to women interested in metabolic health. The sensor is a third-party Abbott device; Hello Inside is a coaching layer. Whether the subscription fee on top of the sensor is worth it depends on how much app guidance you value. We mention it neutrally, not as a recommendation.

A recurring confusion: DiGA-approved digital health apps in Germany (Una Health, Glucura) are reimbursed for diagnosed type-2 diabetes, not for healthy users. Do not assume insurance coverage transfers to wellness use.

For a healthy adult in DACH 2026, the realistic Selbstzahler path: buy two Libre 3 sensors at an Apotheke (about EUR 120 to 150 for one month), use the manufacturer's free app, optionally layer a coaching service. No reimbursement applies.

If your goal is an evidence-based metabolic check-in, the established biomarkers are unglamorous, cheap and well-validated. Fasting glucose plus HbA1c gives you a calibrated read on average glycemia over 8 to 12 weeks. Fasting insulin and the derived HOMA-IR score is the best non-invasive proxy for insulin resistance available to a Hausarzt. A 75 g OGTT with 0, 60 and 120-minute glucose and insulin measurements is the gold standard for catching impaired glucose tolerance and early hyperinsulinemia before HbA1c moves.

For cardiovascular and longevity-adjacent biomarkers: ApoB or LDL particle number, Lp(a) at least once in a lifetime, triglycerides, hs-CRP and HDL-C. Combined with blood pressure, waist circumference and resting heart rate, these have decades of randomized-trial evidence. CGM-derived glucose variability has none of this evidence in healthy adults.

The pragmatic comparison: in DACH 2026, a comprehensive private Selbstzahler blood panel covering HbA1c, fasting glucose, fasting insulin, lipid panel including ApoB, hs-CRP and basic thyroid markers costs EUR 150 to 350 once per year. One year of CGM sensors as a self-pay healthy user costs EUR 700 to 1500. The blood panel has decades of outcome data behind it. The CGM has Stanford glucotypes and an n=153 healthy-cohort baseline. Both can be useful. Only one is validated.

The right way to think about CGM for a curious longevity-minded adult: a structured two-week experiment, once, to see what your own meals and habits do to your curve — knowing that the result is data, not destiny. Then go back to the validated biomarkers and a doctor who actually reads them.