CGM Rate-of-Change Interpretation: What Your Arrows Actually Mean

What Does CGM Rate of Change Actually Measure?

CGM rate of change (ROC) quantifies how rapidly interstitial glucose is shifting, expressed in mg/dL per minute. Most devices sample every 1 to 5 minutes and calculate a rolling delta over the prior 15 to 20 minutes. That number then maps to an arrow symbol displayed on the receiver or app.

The Arrow Scale Most Devices Use

Dexcom G6 and G7, the most widely studied CGM platforms, use a five-tier arrow system derived from the 15-minute glucose trend [1]:

• Horizontal arrow: changing <1 mg/dL per minute
• Single angled up or down arrow: changing 1 to 2 mg/dL per minute
• Double angled up or down arrow: changing 2 to 3 mg/dL per minute
• Single vertical (straight up or down) arrow: changing >3 mg/dL per minute

The Abbott FreeStyle Libre 3 uses a similar scale, though it computes the trend over a slightly different window [2]. Knowing the exact thresholds your device uses matters. A Dexcom double-up arrow and a Libre double-up arrow both mean "fast rise," but the numeric cutoffs differ by roughly 0.5 mg/dL per minute.

Why Interstitial Lag Changes Arrow Reliability

Interstitial fluid glucose lags behind blood glucose by approximately 5 to 15 minutes, a delay confirmed in a 2013 analysis published in Diabetes Technology and Therapeutics [3]. During rapid glucose change, that lag means the CGM number is a rear-view mirror. The arrow, by contrast, is forward-looking. When glucose is rising quickly, the actual blood glucose is already higher than what the screen shows. A double-up arrow at a displayed 140 mg/dL may reflect a true capillary glucose already above 160 mg/dL.

Accuracy Limits Worth Knowing

CGM mean absolute relative difference (MARD) for the Dexcom G7 is 8.2% in clinical testing submitted to the FDA [4]. That margin of error is acceptable at 100 mg/dL but becomes clinically significant near hypoglycemic thresholds. Always confirm a <70 mg/dL reading with a fingerstick before treating if the patient is asymptomatic and the arrow is flat or slowly descending.

The ADA and ATTD Consensus Targets You Should Know

The 2024 American Diabetes Association Standards of Care and the 2023 Advanced Technologies and Treatments for Diabetes (ATTD) International Consensus guidelines both define standardized CGM metrics [5][6]. Clinicians and patients need to use the same reference frame.

Time in Range (TIR)

For most adults with type 1 or type 2 diabetes, the TIR target is more than 70% of readings between 70 to 180 mg/dL. The ATTD 2023 consensus states: "Time in range of 70 to 180 mg/dL greater than 70% is associated with reduced microvascular complication risk and is the primary CGM outcome target for most adults with diabetes" [6].

Each 5% improvement in TIR (roughly 72 minutes per day) correlates with a 0.25 percentage-point reduction in HbA1c, based on regression data from the landmark DIaMonD trial [7].

Time Below Range (TBR)

Time below 70 mg/dL should stay under 4% of readings. Time below 54 mg/dL should stay under 1%. These thresholds were validated in the GOLD randomized crossover trial (N=161), which showed that CGM-guided management significantly reduced hypoglycemic events versus self-monitored blood glucose [8].

Time Above Range (TAR)

Time above 180 mg/dL should stay under 25% of readings. For older adults or those with high hypoglycemia risk, the ADA 2024 guidance accepts TIR above 50% as a reasonable secondary target [5].

Glycemic Variability Metrics

Coefficient of variation (CV) below 36% is the accepted threshold for stable glycemic control, per the ATTD consensus [6]. CV above 36% signals excessive variability independent of mean glucose, and high variability is independently associated with hypoglycemia risk and cardiovascular outcomes in type 1 diabetes [9].

How to Use Trend Arrows in Clinical Decision-Making

A CGM number without its arrow is like a speedometer reading without knowing whether the driver is accelerating or braking. The arrow modifies every treatment decision.

Insulin Dosing Adjustments

The most widely cited practical framework comes from Aleppo et al., published in the Journal of Diabetes Science and Technology in 2017 [10]. That paper proposed arrow-based correction factors for rapid-acting insulin:

• Horizontal arrow: use standard correction dose
• Single diagonal up: add 10 to 20% to the calculated dose
• Double diagonal up: add 20 to 30%
• Straight-up arrow: add 30 to 40% and recheck in 20 minutes
• Single diagonal down: reduce dose by 10 to 20%
• Double diagonal down: reduce dose by 20 to 30% or hold correction entirely

These percentages are starting points. Individual insulin sensitivity and carbohydrate ratio adjustments still apply. No fixed rule replaces the judgment of a clinician who knows the specific patient's history.

Carbohydrate Intake During Exercise

Glucose can drop more than 3 mg/dL per minute during sustained aerobic exercise in insulin-treated individuals, a rate documented in a study published in Diabetologia [11]. A double-down arrow during moderate-intensity exercise at 90 mg/dL means the patient may reach 60 mg/dL within 10 minutes. Consuming 15 to 20 g of fast-acting carbohydrate and rechecking in 15 minutes is the standard recommendation from the ADA exercise guidelines [5].

Overnight Monitoring and Alarm Settings

Setting a CGM low alert at 80 mg/dL (rather than the default 70 mg/dL) gives an extra 10 to 15 minutes of warning when glucose is falling at 1 to 2 mg/dL per minute. The T1D Exchange Quality Improvement Collaborative data show that patients who use predictive low alerts reduce time below 70 mg/dL by approximately 30 minutes per day compared to threshold-only alarms [12].

Optimal CGM Values for Non-Diabetic and Longevity-Focused Users

CGM use has expanded well beyond diabetes management. Metabolically healthy individuals, people on GLP-1 agonists, and those pursuing longevity optimization increasingly wear CGMs. Their targets differ from ADA diabetes thresholds.

Glucose Ranges in Metabolically Healthy Adults

A 2019 study by Danne et al. And subsequent data from the Levels Health internal cohort show that metabolically healthy adults without diabetes typically maintain glucose between 70 to 120 mg/dL for more than 90% of readings, with post-meal peaks rarely exceeding 140 mg/dL [13]. The 180 mg/dL upper bound of the standard TIR range is a diabetes-management target, not an optimal health target.

Post-Meal Excursion Targets

Glucose rising more than 30 mg/dL above fasting within 30 minutes of eating reflects rapid carbohydrate absorption. A single or double diagonal-up arrow at that moment indicates an excursion trajectory. Some longevity-medicine practitioners use a post-meal peak below 140 mg/dL as a personal target, though no randomized trial has shown that reducing post-meal peaks below 140 mg/dL in non-diabetic individuals reduces hard outcomes. The data to support sub-140 mg/dL as a mortality target in healthy populations does not yet exist.

Fasting and Overnight Glucose Stability

Overnight glucose in metabolically healthy adults tends to hold within a 15 to 20 mg/dL band, with slow, gradual drift. Wide overnight variability, defined as swings exceeding 40 mg/dL, warrants evaluation for nocturnal hypoglycemia, dawn phenomenon, or Somogyi effect, even in non-diabetic users.

CGM Metrics for GLP-1 and TRT Patients

Patients on semaglutide (Ozempic, Wegovy) or tirzepatide (Mounjaro, Zepbound) often see TIR improve substantially within 4 to 8 weeks of dose titration. In the SURPASS-2 trial (N=1,879), tirzepatide 15 mg reduced mean glucose from 176 mg/dL to approximately 126 mg/dL at 40 weeks, with TIR improving from roughly 43% to over 80% in participants who used CGM [14]. Testosterone replacement therapy in hypogonadal men with type 2 diabetes has shown modest reductions in fasting glucose and HbA1c, as documented in the TRAVERSE trial sub-analysis [15], though the magnitude of CGM metric improvement from TRT alone is not yet well-characterized.

Reading Glycemic Variability: Beyond the Arrow

The real-time arrow tells you what glucose is doing right now. Variability metrics summarize what it has been doing across days and weeks, and they predict future risk.

Mean Amplitude of Glycemic Excursion (MAGE)

MAGE quantifies the average size of significant glucose excursions, filtering out minor noise. A MAGE above one standard deviation of the mean glucose indicates high variability. High MAGE correlates with oxidative stress markers and endothelial dysfunction in studies published in Diabetes Care [16]. Reducing MAGE requires identifying which meals, activities, or stress events trigger the largest spikes.

Glucose Management Indicator (GMI)

GMI is calculated from the formula: GMI (%) = 3.31 + 0.02392 x mean glucose (mg/dL). It estimates what HbA1c would be based on CGM data alone, and is most accurate when derived from at least 14 days of CGM wear [17]. GMI diverges from laboratory HbA1c by more than 0.5 percentage points in patients with hemoglobin variants, iron deficiency, or chronic kidney disease.

The AGP Report

The Ambulatory Glucose Profile (AGP) report consolidates 14+ days of CGM data into a single-page visualization. It shows the median glucose trace, the interquartile range (25th, 75th percentile), and the 10th, 90th percentile band across a 24-hour overlay. The International Diabetes Center and ATTD consensus both recommend AGP as the standard reporting format for all CGM downloads [6].

Common Interpretation Mistakes and How to Avoid Them

Treating the Number Without Reading the Arrow

Administering a correction insulin dose based on a 160 mg/dL reading and a double-down arrow can produce dangerous hypoglycemia within 30 to 45 minutes. The glucose is already falling rapidly. The arrow should delay or reduce the correction.

Calibration and Sensor Placement Errors

CGM accuracy degrades within the first 12 to 24 hours after sensor insertion (the "first-day effect") and near the end of sensor life. Sensors placed over scar tissue or lipohypertrophy report less accurate values. The FDA cleared the Dexcom G7 and Abbott FreeStyle Libre 3 as factory-calibrated, factory-calibrated devices that do not require routine fingerstick calibration under normal conditions [4][2]. Calibration entry during rapid glucose change introduces error, not correction.

Alarm Fatigue

Patients who set alerts at both 70 mg/dL low and 180 mg/dL high report alarm fatigue, and many disable alerts entirely within 60 days of starting CGM. A 2020 survey published in Diabetes Technology and Therapeutics found that 53% of CGM users had disabled at least one alert within the first three months [18]. Narrowing alert range during the day and widening it overnight, based on personal patterns, reduces fatigue without sacrificing safety.

Confusing CGM Glucose with Plasma Glucose

CGM reports interstitial glucose. Plasma glucose from a laboratory venous draw is the reference standard. CGM values will differ from simultaneous plasma values by 10 to 15% even on well-functioning sensors, and the discrepancy widens during rapid change. Diagnosing hypoglycemia or hyperglycemia for clinical documentation requires a laboratory or cleared point-of-care plasma glucose value, not a CGM reading alone.

How to Set Up a CGM Review Visit

A structured CGM download review visit focuses on four questions: What is the TIR? Where does glucose fall below 70 mg/dL? What times of day show the highest variability? Does the GMI match the laboratory HbA1c?

The 14-Day Download Minimum

Fewer than 14 days of CGM data produce unreliable AGP statistics. The ATTD consensus requires a minimum of 70% data capture over 14 days for any metric to be reported as valid [6]. A patient who wore a sensor for 10 days but had 2 days of missing data due to calibration errors should be asked to complete another wear cycle before interpreting CV or MAGE.

Structured Meal and Activity Logging

CGM data without context is hard to interpret. Pairing CGM with a 3-day food log and an exercise log allows the clinician to identify the specific foods or activities driving excursions. Apps that sync meal logs with CGM traces, such as the Dexcom Clarity and Abbott LibreView platforms, reduce the time needed for pattern review at the visit.

When to Order a Simultaneous HbA1c

Order a laboratory HbA1c alongside CGM download review for any patient where GMI and clinical expectation diverge by more than 0.5 percentage points. Hemolytic anemia, thalassemia trait, and recent blood transfusion all falsely lower HbA1c while leaving CGM-derived GMI unaffected, making GMI a more reliable glycemic index in those populations [17].