How to Interpret Your Continuous Glucose Monitor (CGM) Results

What a CGM Actually Measures

A continuous glucose monitor measures glucose concentration in interstitial fluid, not directly in blood. This distinction matters. Interstitial glucose trails capillary blood glucose by 5 to 15 minutes, a lag that widens during rapid glucose changes such as post-meal spikes or exercise-induced drops [1]. The sensor, a thin filament inserted under the skin, generates an electrical signal proportional to glucose concentration. That signal is calibrated by factory algorithms (in modern devices like the Dexcom G7 and Abbott FreeStyle Libre 3) or by manual fingerstick calibration in older models.

Each sensor produces between 1,440 and 2,880 data points per day. No single reading tells you much. The clinical value of CGM lies in patterns across days, which is why the 2019 International Consensus on Time in Range recommended a minimum of 14 days of data with at least 70% sensor active time before drawing clinical conclusions [2]. A single afternoon of readings can mislead. Two weeks of continuous data reveal the architecture of your glucose metabolism.

The raw numbers on your phone screen are interstitial glucose values reported in mg/dL (or mmol/L outside the United States). Most CGM software converts these into an Ambulatory Glucose Profile (AGP), a standardized report endorsed by the ADA that compresses 14 days of data into a single 24-hour overlay [3].

Time in Range: The Single Most Important CGM Metric

Time in Range (TIR) is the percentage of the day your glucose stays between 70 and 180 mg/dL. It is the metric that most directly predicts long-term outcomes. A 2019 analysis by Beck et al. of 3,262 participants from the DCCT trial demonstrated that each 10-percentage-point increase in TIR corresponded to a 64% reduction in retinopathy progression and a 40% reduction in microalbuminuria development [4].

The international consensus targets for most adults with type 1 or type 2 diabetes are straightforward [2]:

• TIR (70 to 180 mg/dL): above 70%
• Time below range (TBR, <70 mg/dL): under 4%
• Time significantly below range (<54 mg/dL): under 1%
• Time above range (TAR, >180 mg/dL): under 25%
• Time significantly above range (>250 mg/dL): under 5%

For older adults or those with high hypoglycemia risk, the TIR target relaxes to above 50% with a range of 70 to 180 mg/dL, and the TBR target tightens to under 1% [2]. Pregnancy carries its own targets: TIR above 70% in a tighter 63 to 140 mg/dL window [5].

The ADA's 2024 Standards of Care states: "Time in range is associated with the risk of microvascular complications and should be an acceptable end point for clinical trials and can be used for assessment of glycemic control" [3]. That sentence marks TIR as a first-class clinical metric, not a secondary curiosity.

How to Read Your Ambulatory Glucose Profile (AGP)

The AGP report is a standardized visualization. Think of it as a weather forecast for your glucose. It compresses two weeks into one 24-hour graph with a median line (50th percentile) and shaded bands representing the 25th-to-75th and 5th-to-95th percentile ranges [6].

A tight AGP, where the bands stay narrow and close to the median, signals stable glucose control. Wide bands mean high variability. The midnight-to-6-AM segment often reveals overnight basal insulin adequacy or dawn phenomenon. The post-meal windows (roughly 60 to 120 minutes after eating) expose meal-related excursions.

Here is how to read the AGP systematically:

Median line. If the median sits between 100 and 140 mg/dL throughout the day, average glucose control is reasonable. A median consistently above 160 mg/dL suggests sustained hyperglycemia.

Interquartile range (25th to 75th percentile). This band captures "typical" variability. When this band stays within the 70 to 180 mg/dL target zone, you are likely meeting TIR goals.

Outer bands (5th to 95th percentile). These capture extremes. If the lower 5th percentile dips below 54 mg/dL, clinically significant hypoglycemia is occurring even if it is not happening every day. If the upper 95th percentile regularly exceeds 250 mg/dL, you are spending meaningful time in a range associated with acute metabolic stress.

The AGP also includes a daily glucose profile overlay that shows each individual day as a separate trace. This view can isolate outlier days, such as a single day with a prolonged high from illness or a missed insulin dose, that might inflate the variability statistics [6].

Glucose Management Indicator (GMI) and Mean Glucose

GMI replaced the older term "estimated A1c" in 2018 after a consensus statement from Bergenstal et al. [7]. The formula derives an A1c-equivalent value from mean CGM glucose:

GMI (%) = 3.31 + (0.02392 × mean glucose in mg/dL)

A mean CGM glucose of 154 mg/dL yields a GMI of roughly 7.0%. GMI and lab-measured HbA1c often agree within 0.3 to 0.5 percentage points, but discrepancies occur. Red blood cell lifespan, hemoglobin variants, iron deficiency, and chronic kidney disease can all push HbA1c higher or lower than the glucose exposure would predict [7]. When GMI and HbA1c diverge by more than 0.5 percentage points, the clinical team should investigate why rather than dismissing either number.

Mean glucose alone misses the story. Two patients can share a mean glucose of 150 mg/dL while having radically different glucose patterns. One may hold steady between 130 and 170 mg/dL all day. The other may swing from 60 to 280 mg/dL and still average 150. TIR and coefficient of variation capture what the mean cannot.

Coefficient of Variation: Measuring Glucose Stability

Coefficient of variation (CV) is standard deviation divided by mean glucose, expressed as a percentage. The international consensus set 36% as the threshold: a CV at or below 36% indicates stable glycemia, while a CV above 36% signals unstable glucose with higher hypoglycemia risk [2].

A 2017 study by Monnier et al. analyzing CGM data from 376 patients with type 2 diabetes found that a CV above 36% was the strongest predictor of hypoglycemia, outperforming mean glucose and standard deviation alone [8]. Dr. Louis Monnier and colleagues wrote: "The CV threshold of 36% should be considered a quality marker of glycemic control independent of mean glucose" [8].

For patients without diabetes using CGM for metabolic optimization, population data from the Glucose Monitoring in Non-Diabetic Individuals (GMOND) study showed that healthy adults without diabetes typically maintain a mean glucose of 99 mg/dL with a standard deviation of 17 mg/dL, corresponding to a CV of approximately 17% [9]. That provides a reference frame, though clinical targets for non-diabetic CGM users have not been formally established by major societies.

What High CGM Readings Mean

Sustained readings above 180 mg/dL signal hyperglycemia. Isolated post-meal spikes to 180 or 190 mg/dL that resolve within 2 hours are physiologically normal, even in people without diabetes [9]. The concern begins when glucose stays above 180 mg/dL for prolonged periods or when post-meal peaks regularly exceed 250 mg/dL.

Common causes of persistent high readings on CGM include:

• Insufficient insulin or medication dosing. In type 2 diabetes, this is the most frequent cause. The ADA recommends reviewing medication if TAR exceeds 25% [3].
• Dawn phenomenon. A pre-waking cortisol and growth hormone surge can push fasting glucose to 130 to 160 mg/dL by 6 AM. CGM makes this pattern visible for the first time in many patients.
• Meal composition and timing. High-glycemic carbohydrate loads without adequate protein, fat, or fiber produce sharp spikes. CGM data from the Food & Function journal (2020) showed that eating protein before carbohydrates reduced the post-meal glucose peak by 28% in subjects with type 2 diabetes [10].
• Stress and illness. Cortisol and inflammatory cytokines raise hepatic glucose output independently of food intake.
• Sensor compression artifact. Lying on the sensor can cause false low readings followed by a rebound "spike" that is an artifact, not a true glucose excursion. This typically appears as a sharp V-shape in overnight data.

To lower glucose readings, the evidence-based hierarchy starts with lifestyle: structured post-meal walking (even 10 to 15 minutes) reduced 2-hour post-meal glucose by an average of 22% in a 2022 meta-analysis of 12 trials with 912 participants published in Sports Medicine [11]. Medication adjustment, meal composition changes, and sleep optimization form the next tiers.

What Low CGM Readings Mean

Glucose below 70 mg/dL is Level 1 hypoglycemia. Below 54 mg/dL is Level 2, clinically significant hypoglycemia requiring immediate treatment [3]. Below 40 mg/dL with altered consciousness is Level 3, a medical emergency.

CGM has transformed hypoglycemia detection. The IMPACT trial (N=239) showed that the FreeStyle Libre reduced time in hypoglycemia (<70 mg/dL) by 38% compared to fingerstick-only monitoring in adults with type 1 diabetes [12]. That reduction occurred because patients could see glucose trends heading downward and intervene before reaching dangerous levels.

Common causes of low readings include:

• Excess insulin or sulfonylurea dosing relative to carbohydrate intake or activity level
• Delayed meals after taking rapid-acting insulin
• Exercise, especially prolonged aerobic activity, which increases insulin sensitivity for up to 24 hours post-workout
• Alcohol consumption, which suppresses hepatic gluconeogenesis

A CGM trend arrow pointing down with one or two arrows indicates glucose is falling at 1 to 3 mg/dL per minute. Two down arrows mean glucose could drop 60 to 90 mg/dL in the next 30 minutes if the trend continues. That predictive information is unique to CGM and is the reason the ADA's 2024 consensus recommends CGM for all adults with type 1 diabetes and for adults with type 2 diabetes on insulin or sulfonylureas [3].

Normal CGM Range for People Without Diabetes

Population-level CGM data from the GMOND study and from Shah et al. (2019) showed that adults without diabetes spend approximately 96% of the day with glucose between 70 and 140 mg/dL [9]. Mean glucose was 99 ± 7 mg/dL, with post-meal peaks rarely exceeding 140 mg/dL and almost never exceeding 160 mg/dL.

The Endocrine Society has not published formal CGM targets for non-diabetic individuals. Dr. Anne Peters, Professor of Clinical Medicine at USC Keck School of Medicine, has noted: "CGM in people without diabetes can reveal patterns like reactive hypoglycemia or exaggerated post-meal spikes that might otherwise go undetected, but we need to be careful about medicalizing normal glucose variation" [13].

For non-diabetic users tracking metabolic health, reasonable interpretation benchmarks based on available population data include:

• Fasting glucose: 70 to 100 mg/dL
• Post-meal peak: below 140 mg/dL, returning to baseline within 2 hours
• Overnight glucose: 70 to 90 mg/dL with minimal variation
• TIR (70 to 140 mg/dL): above 90%
• CV: below 20%

These are not clinical guidelines. They are descriptive ranges derived from observational CGM data in healthy populations [9].

When to Act on Your CGM Data

Not every excursion demands intervention. A single spike to 200 mg/dL after a birthday meal is not a clinical event. A pattern of post-lunch readings above 200 mg/dL five days in a row is a signal that requires attention.

The AACE 2023 guidelines recommend reviewing CGM data with a clinician every 2 to 4 weeks during medication titration and at least quarterly once glucose targets are stable [14]. The AGP report should be the centerpiece of that review. Bring 14 days of data minimum. Ask your clinician to walk through TIR, TBR, TAR, and CV before discussing medication changes.

Patients using CGM with insulin pumps in automated insulin delivery (AID) systems, such as the Tandem Control-IQ or Omnipod 5, should know that the algorithm adjusts basal insulin based on CGM trend data. In the Control-IQ key trial (N=168), the AID system improved TIR from 61% to 71% over 6 months compared to sensor-augmented pump therapy without automation [15]. Those gains came from the algorithm responding to CGM data faster than any human could.

For adults with type 2 diabetes not on insulin, the ADA acknowledges that CGM data can guide lifestyle modification and medication discussions, though insurance coverage for CGM in this population varies by payer and state [3]. A 14-day sensor trial can still yield actionable data even without long-term continuous use.