How Is GMI Calculated From My CGM Average Glucose
GMI is 3.31 + 0.02392 × your average glucose in mg/dL. It is an A1C-like estimate from sensor data, not a lab measurement.
If your CGM export shows an average glucose of 150 mg/dL, the GMI on the report is 3.31 + 0.02392 × 150, which is 6.9%. GMI stands for Glucose Management Indicator. It is an A1C-like percentage calculated from your sensor average, not a lab result.
GMI = 3.31 + 0.02392 × average glucose in mg/dL. It is an estimate, not a lab A1C.
How is GMI calculated from my CGM average glucose?
Take the mean of all sensor readings in mg/dL for the period you are looking at. Multiply that mean by 0.02392. Add 3.31. The result is your GMI as a percentage.
| Average glucose (mg/dL) | GMI (%) |
|---|---|
| 100 | 5.7 |
| 120 | 6.2 |
| 140 | 6.7 |
| 150 | 6.9 |
| 180 | 7.6 |
GMI from the formula 3.31 + 0.02392 × average glucose in mg/dL.
If your export is in mmol/L, multiply the average by 18.018 first, or use GMI = 3.31 + 0.431 × average glucose in mmol/L. The formula is a straight line, so a higher average always produces a higher GMI. For context, lab A1C cutoffs are 5.7% for prediabetes and 6.5% for diabetes; GMI is not a diagnostic test. A GMI below 5.7% corresponds to an average glucose around 100 mg/dL or lower.
Why your sensor report shows an A1C-like number at all
Lab A1C measures how much glucose has attached to hemoglobin inside red blood cells. Although red blood cells live about 4 months, A1C reflects a 2 to 3 month weighted glucose window, with recent weeks counting more heavily. A CGM reads interstitial fluid every few minutes, usually for one or two weeks. That is a much shorter window.
The GMI formula came from a regression by Bergenstal and colleagues in 2018. Researchers compared mean CGM glucose with measured lab A1C in a large dataset of people with diabetes and fitted the relationship. The formula is that line. It was named Glucose Management Indicator partly to avoid calling it an estimated A1C, because it is not a direct measurement of glycated hemoglobin.
When is GMI reliable?
The number on a three-day sensor is not reliable. GMI needs enough data to smooth out meals, sleep, sensor noise and the first-day warm-up. The first day of a new sensor is the least stable because the filament is still settling into tissue. If you just inserted a sensor, wait before taking the average seriously.
Practical rules: - Use at least 14 days of sensor data for a stable GMI. - Look for at least 70% sensor coverage over that period as a common CGM data-quality target in the International Consensus on Time in Range, not an absolute rule. - Do not make a decision from a single day. - Treat sharp overnight dips that recover fast as possible compression lows, not real glucose drops.
Sensor gaps reduce reliability, but they do not by themselves drag the average down. Missing data can shift the mean up or down depending on which hours were missed. A sensor that records low nighttime readings can lower the average if those readings are included and are not offset by high readings; one that records high readings can push it up.
Why is my GMI different from my lab A1C?
That is normal. GMI and lab A1C are different measurements. They are correlated, but individual agreement varies, and the difference can be large enough to matter in some people.
What changes the gap: - CGM reads interstitial fluid, not venous blood. - CGM average covers only the days you actually wore the sensor. - A1C depends on red blood cell survival. Iron deficiency, B12 or folate deficiency, hemoglobin variants, blood donation, pregnancy and kidney disease can all shift A1C without changing your average glucose. - Sensor bias and calibration can move the CGM average.
Recent changes in glucose control show up in GMI before A1C, because A1C reflects a 2 to 3 month weighted glucose window that changes slowly. If your GMI says 7.0% and your lab A1C says 6.2%, the lab is not wrong and the sensor is not broken. They are estimating related but different things.
What the average leaves out
GMI is built from a mean. A mean hides variability. Two people can have the same average glucose and very different glucose patterns: one nearly flat, the other swinging from 70 to 180 after meals. The second person may have a normal GMI and still have large post-meal rises worth understanding. That is why the same CGM export should be read alongside time in range, variability and the shape of the day.
If you already have a sensor, drop your LibreView, Dexcom Clarity, Stelo, Lingo or SiBionics export at the free CGM report and see your GMI, AGP, time in range and biggest rises in the browser, no account. If you want a sensor, HealthOS Sensor is a month of continuous glucose with the app for $70, no membership, outside the US.
Written by
HealthOS Research