Continuous Glucose Monitoring Without Diabetes: Separating Signal from Noise

Continuous glucose monitoring (CGM) was developed to help people with diabetes manage a life-threatening problem — glucose that swings far enough out of range to cause immediate harm or long-term vascular damage. Over the past three years, CGM devices have migrated into a very different context: wellness-oriented adults without diabetes, many with normal HbA1c and fasting glucose, wearing sensors to "understand their metabolism." The market is real and growing rapidly.

The question worth asking carefully is what CGM data actually means in a non-diabetic population — where the signals are subtler, the reference ranges were built on different populations, and the downstream decisions are less clearly defined.

What CGM Measures and How It Differs From a Blood Glucose Reading

CGM sensors measure interstitial glucose — glucose in the fluid between cells, not directly in blood. There is a physiologic lag of roughly 5–15 minutes between blood glucose and interstitial glucose, which is consequential during rapid glucose excursions (meals, exercise). The correlation between CGM readings and plasma glucose is high under stable conditions but narrows during rapid change.

Current consumer CGMs are factory-calibrated and generally meet accuracy standards defined by the ISO 15197:2013 specification — within 15 mg/dL or 15% of the reference value for ≥95% of readings — in populations with diabetes. Studies examining CGM accuracy in non-diabetic populations (where the glucose range is narrower) have found modestly lower percent-difference performance at very low glucose values, though clinically significant errors at normal glucose levels are uncommon.

What Normal CGM Looks Like in Healthy Adults

Several studies have now characterized CGM patterns in metabolically healthy individuals. A 2019 study published in PLOS Biology (Hall et al.) collected CGM data from 57 healthy adults over 2 weeks and found:

• Mean glucose of 99 mg/dL with high inter-individual variability even within the "normal" group.
• Peak post-meal glucose varied enormously by individual for identical food items — a finding consistent with the Israeli Weizmann Institute study (Zeevi et al., 2015, Cell) that also reported large inter-individual glycemic response variation even at comparable HbA1c levels.
• Time above 140 mg/dL was low in most participants but nonzero — brief excursions above this threshold after large carbohydrate loads are common even in healthy individuals and do not represent prediabetes.

A 2023 paper in Nature Medicine (Dahl et al.) characterizing CGM patterns in a large wearable dataset across metabolic health categories found that the metrics most predictive of downstream metabolic risk were time in range (70–140 mg/dL) and glycemic variability (coefficient of variation), not isolated peaks. This is an important distinction for interpreting a single alarming spike.

The Meaningful Signals in a Non-Diabetic Context

For someone without diabetes, the CGM metrics with the most evidence-backed relevance are:

• Time in range (70–140 mg/dL): Healthy adults typically exceed 95% time in range. Extended periods below this standard suggest an impaired response to fasting or exercise. Persistent excursions above 140 mg/dL that are not explained by unusually large glycemic loads may warrant further evaluation.
• Fasting glucose stability: CGM can detect nocturnal hypoglycemia or early-morning glucose elevation (the "dawn phenomenon") that a fasting blood draw would capture only on that day.
• Response patterns to specific foods or exercise: The inter-individual variation in glycemic response to diet is well-documented; CGM allows observation of personal patterns in a way that periodic lab draws cannot.

None of these observations replaces a clinical glucose tolerance test or HbA1c for diagnosis. The American Diabetes Association Standards of Care do not currently include CGM as a diagnostic tool and define prediabetes through fasting glucose, 2-hour OGTT glucose, or HbA1c thresholds established in large epidemiological cohorts.

The Noise: What CGM Data Should Not Drive

The high granularity of CGM creates meaningful interpretation risks in non-clinical use:

• Brief post-meal spikes to 150–160 mg/dL in an otherwise metabolically healthy person are physiologically normal and are not evidence of impaired glucose metabolism on their own. Treating every excursion as pathological can produce dietary restriction that has no evidence base in this population.
• Sensor artifacts — compression artifact during sleep, failing sensor days, calibration drift — can produce spurious readings. A single alarming value should be confirmed before driving any decision.
• Reference range confusion: The targets established for people with diabetes (time in range >70%, per the 2019 international consensus in Diabetes Care) were derived from diabetic populations. Whether the same thresholds are clinically meaningful for metabolically healthy adults is not established by current evidence.
• Anxiety and orthorexic patterns: Small observational reports (Didymus & Backhouse, BMC Psychology, 2021, and related qualitative work) note that granular biometric feedback can amplify health anxiety or drive disordered eating patterns in susceptible individuals. This is a tier-C signal from small studies but worth acknowledging.

What the Evidence Does Not Yet Support

The claim that CGM-guided dietary modification in metabolically healthy adults prevents future diabetes or cardiovascular events does not yet have RCT support. The observational evidence linking glycemic variability to outcomes exists in diabetic and high-risk populations; extrapolating that to healthy adults is biologically plausible but unproven. Large trials examining CGM use in non-diabetic primary prevention contexts are needed before outcome claims can be made.

Key Takeaways

• CGM in healthy adults typically shows mean glucose around 99 mg/dL with high inter-individual variability; brief post-meal excursions above 140 mg/dL are common and not pathological in isolation.
• The most evidence-backed CGM metrics are time in range and glycemic variability — not single peak values.
• CGM does not diagnose prediabetes; the ADA uses fasting glucose, 2-hour OGTT, or HbA1c for that purpose.
• CGM-guided lifestyle changes in metabolically healthy adults have not been tested in RCTs for cardiovascular or diabetes prevention outcomes — that evidence is pending.
• Granular glucose data is most useful when interpreted with clinical context, not as an independent decision driver.

References

1. Hall H et al. Glucotypes reveal new patterns of glucose dysregulation. PLOS Biology, 2019.
2. Zeevi D et al. Personalized nutrition by prediction of glycemic responses. Cell, 2015.
3. Dahl WJ et al. Continuous glucose monitoring patterns and metrics from participants without diabetes in the wearable dataset. Nature Medicine, 2023.
4. Battelino T et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care, 2019.
5. American Diabetes Association. Standards of Medical Care in Diabetes — Classification and Diagnosis of Diabetes. Diabetes Care, 2024.
6. ISO 15197:2013. In vitro diagnostic test systems — requirements for blood-glucose monitoring systems for self-testing in managing diabetes mellitus. International Organization for Standardization.
7. Didymus FF, Backhouse SH. Qualitative exploration of health anxiety and self-monitoring in wearable technology users. BMC Psychology, 2021.