CGM Sensors: Continuous Glucose Monitoring Guide

TL;DR: Commonly cited reference values for healthy adults: fasting 70–90 mg/dl (3.9–5.0 mmol/l), postprandial peak below 140 mg/dl (7.8 mmol/l), Time in Range above 90 %, average glucose below 105 mg/dl. Sensors like FreeStyle Libre 3 and Dexcom G7 run 10–14 days. A CGM shows what HbA1c cannot — but healthy people do not need it permanently.

This article does not replace medical advice. Whether glucose monitoring makes sense, how values are interpreted and which measures or treatments follow from them should be clarified with a physician. If you regularly see glucose values below 55 mg/dl or above 180 mg/dl, please seek medical advice.

What a CGM Actually Measures

A Continuous Glucose Monitor measures glucose — but not in blood. The sensor sits with a thin filament a few millimeters below the skin and measures glucose in the interstitial fluid, the space between cells. Every 1 to 5 minutes it delivers a reading. After 10 to 14 days the sensor is replaced.

The difference to blood glucose is real. When blood sugar rises quickly after a meal, glucose arrives in the interstitium with a 5 to 15 minute delay. Same in reverse when it drops. At stable values the difference is minimal, at rapid changes it matters.

Accuracy is expressed as MARD — Mean Absolute Relative Difference to lab blood glucose. FreeStyle Libre 3 and Dexcom G7 sit around 8 to 9 percent. Good enough for trends, too imprecise for insulin therapy decisions in type 1 diabetes without blood-glucose verification.

A practical example: after a plate of rice, capillary blood glucose shows 165 mg/dl at 45 minutes. The CGM shows 148 mg/dl — and reaches 165 mg/dl only at 55 minutes. Both readings are correct. They simply measure different compartments.

Brief History: From Diabetes Tool to Biohacker Gadget

CGM was developed for type 1 diabetics. The first commercial systems (MiniMed) launched in 1999. For years they were expensive, inaccurate, and prescription-only.

The turning point came in 2014 with Abbott’s FreeStyle Libre. No fingerstick calibration, simple application, lower price. From around 2020 onwards, biohackers and health-optimizers discovered the device. Companies like Levels, Nutrisense and Veri began marketing CGM to non-diabetics — as a tool for observing individual nutrition responses.

In 2024, Stelo (Dexcom) became the first OTC CGM in the US specifically for non-diabetics. Lingo (Abbott) followed shortly after. In Europe the rollout is slower.

In parallel, studies like PREDICT (Zeevi 2015, Berry 2020) demonstrated that glucose responses to identical meals vary 3 to 10-fold between people. That fueled the biohacker hype further.

Device Overview: Libre, Dexcom, Stelo, Lingo, Eversense

This table compares the key systems by technical features.

Availability: Prescription requirements and access differ by device and country. What applies in an individual case is clarified by the treating physician.

Insurance: Statutory health insurance covers CGM only with a medical indication — typically type 1 diabetes, difficult-to-manage type 2, or gestational diabetes. For biohacking purposes, costs are private.

Reference Values for Healthy Adults

The values cited for non-diabetics are stricter than the official diabetes thresholds. They are not a diagnostic criterion; interpretation is a task for a physician.

The shape of the glucose curve tells more than single readings. A flat curve with gentle rises and slow declines is considered favorable. Sharp spikes followed by reactive lows below 70 mg/dl are considered unfavorable — even with a normal HbA1c.

For deeper understanding of insulin-glucose dynamics, read the guide on insulin resistance and HOMA-IR. How common metabolic syndrome is in Germany is shown in our metabolic syndrome statistics for the DACH region.

What CGM Reveals That HbA1c Cannot

HbA1c is a 3-month average. Two people with identical HbA1c can have completely different glucose profiles. One stays flat at 100 mg/dl all day, the other oscillates between 60 and 180 mg/dl.

Individual food spikes. Rice leads one person to a peak of 180 mg/dl, another only to 120 mg/dl. Oats react the opposite way. Without CGM this cannot be seen.

Dawn phenomenon. Between 3 and 8 AM glucose often rises 10 to 30 mg/dl due to cortisol and growth hormone — without food. This is physiological but can skew fasting readings.

Stress spikes. An important meeting without food can raise glucose by 20 to 40 mg/dl. CGM makes this visible and helps to understand the stress-metabolism link. Combined with HRV tracking the picture becomes more complete.

Nocturnal patterns. Reactive nighttime hypoglycemia, dawn phenomenon, effects of late meals — all of this emerges only with CGM. Many people with poor sleep have nocturnal glucose anomalies invisible without a sensor. More on sleep data in the sleep tracking guide.

Training effects. Strength training briefly raises glucose (cortisol, glucagon). Endurance training lowers it during and for hours after. After intense HIIT, glucose above 160 mg/dl directly after the session is not unusual.

Methodological Pitfalls

CGM is powerful but not perfect. Five sources of error are well known.

Compression artifacts. Sleeping on the sensor compresses interstitial fluid. The sensor then reads falsely low, often showing sudden drops below 60 mg/dl. Waking up without symptoms usually means it was not a real low. Persistently abnormal nighttime values should be assessed by a physician.

Blood-to-interstitium lag. 5 to 15 minutes delay, particularly during rapid changes. When compared with a blood measurement, the CGM reading may be lower — simply because it lags.

Sensor drift. The first 24 hours after insertion are unstable, often running low. The last 24 hours before sensor end become inaccurate too.

Calibration. Libre and Dexcom G7 are factory-calibrated; according to the manufacturers, no manual calibration is needed. At values that do not fit the situation (e.g., 180 mg/dl without a meal), a blood measurement provides clarity.

Meal context. What was eaten before influences the next meal. An isolated “rice test” right after waking shows different values than rice after a workout or after a stressful day. Comparability requires standardized conditions.

For wearable data quality in general, read the guide on wearable data quality.

What Observations and Studies Show

Outside diabetes care, a CGM is usually used for a limited period, for example 2 to 4 weeks. The following questions are frequently examined in studies and self-observation. This is context, not a recommendation.

Food comparisons. The PREDICT studies show that the same foods trigger very different glucose curves in different people. Comparisons usually look at peak and AUC (Area Under the Curve) over the following 2 hours.

Order of food components. Studies describe that the postprandial peak can be lower when protein and fiber are eaten before carbohydrates. The evidence is limited and results vary between individuals.

Movement after eating. Short walks after meals are considered one of the more consistently observed influences on the postprandial curve.

Vinegar. Individual studies examine whether apple cider vinegar before carbohydrate-rich meals influences the glucose curve. The evidence is limited and inconsistent.

Cinnamon, berberine, metformin. Evidence for cinnamon is weak. Berberine is examined in studies for effects on glucose; a CGM merely shows how values behave in an individual case. Metformin is a prescription-only medication whose use is decided exclusively by a physician. This article makes no recommendation to use supplements or medications; intake, dose and possible interactions are clarified by a physician.

Those who record observations usually note timing, portion, previous meal, sleep and training. Without context, CGM data is hard to interpret.

Critical Perspective: Is It Actually Needed?

The uncomfortable truth: most healthy people do not need continuous CGM. Four reasons for restraint.

Cost. 60 to 70 euros per sensor, 4 sensors every 2 months — that is 120 to 150 euros per month without indication. At 1,500 to 1,800 euros per year, asking about benefit is fair.

Pathologizing normal spikes. A peak at 140 mg/dl after pasta is not a disease. Non-diabetics are allowed to spike postprandially. Optimizing every single meal peak can do more harm than good, especially when it leads to restrictive eating.

Orthorexia glucosa. A new phenomenon: people develop compulsive behavior around their glucose curve. No more bread, only salads, fear of fruit. This is not biohacking. It is an eating disorder dressed in tracking. Anyone who recognizes themselves here should seek medical or psychological help.

Weak evidence in healthy people. Randomized trials of CGM in non-diabetics show short-term mild behavioral changes but no robust long-term effects on HbA1c, weight or cardiovascular markers. The biggest benefit lies in the learning curve during the first 4 to 8 weeks.

Restrained use: Proponents describe 1 to 2 sensors per year with a short, intensive observation phase followed by a pause. With increased insulin resistance risk (abdominal fat, family history, elevated HbA1c), a physician decides on frequency and type of measurement. For people who are metabolically healthy, an HbA1c every 12 to 24 months suffices.

Integration Into Lab2go

CGM data unfolds its value only in long-term comparison. Lab2go does not import raw CGM data; individual metrics such as average glucose and Time in Range can be transferred by hand or kept as a note, and compared with HbA1c and fasting glucose in the lab history.

Average glucose vs. HbA1c. A CGM average of 100 mg/dl corresponds to an HbA1c around 5.1 %. The HbA1c lab value can be entered in Lab2go and compared with the CGM average — deviations suggest high variability or lab error.

Glucose vs. sleep. Studies observe stronger glucose responses on the day after short nights. This link is read in the CGM app itself — Lab2go does not correlate it automatically, but it keeps your HbA1c and fasting glucose history over time.

Glucose vs. training. Endurance sessions can lower glucose for a day or two. Weeks with little training tend to show higher averages. This is visible in the CGM app, not in Lab2go.

Glucose vs. weight. In overweight individuals, studies describe an association between weight loss and lower average glucose. Weight and HbA1c can be recorded in the Lab2go lab history; the comparison with CGM data is drawn by the user. The features show all tracking options.

For the methodology of long-term tracking, read the guide on long-term biomarker tracking.

Bottom Line: CGM as a Time-Limited Learning Tool

CGM is not the new mandatory metric for every healthy person. It is a learning tool — most useful as a 2 to 4 week observation, not as permanent measurement.

Three points for context:

1. Define the question. What is to be observed? Meal responses? Sleep effects? Stress signature? Without a question, CGM data is hard to interpret.
2. Clarify with a physician. Whether a sensor makes sense, which values are abnormal and which measures follow is decided by a physician. Intake, dose and treatment with medications or supplements also belong with a physician.
3. Read data in context. Single spikes are not a finding. An HbA1c trend over time provides the more reliable comparison.

Further reading: the biomarker baseline and the comparison of CGM results with HOMA-IR. For platform integration, check the features or compare the pricing.

This article does not replace medical advice. If you repeatedly see values below 55 mg/dl or above 180 mg/dl two hours after meals, please seek medical advice. CGM complements clinical diagnostics — it does not replace them.