What Is the Glucose Management Indicator (GMI)?

The glucose management indicator, or GMI, is the number your continuous glucose monitor’s software shows to estimate what your A1C would be. It’s based only on your sensor’s average glucose. Because it’s an estimate built from sensor readings and not a blood test, it measures something different from the lab A1C your doctor orders. For many people the two won’t match exactly, and a mismatch is usually expected rather than a malfunction.¹ Once you know your own gap between the two, it can tell you and your care team something useful about how your body handles glucose.

How GMI Turns Your Sensor Average Into an A1C-Style Number

A lab A1C measures how much sugar has stuck to the hemoglobin in your red blood cells, which reflects roughly the past two to three months. A CGM (continuous glucose monitor) is a small sensor worn on the skin. It reads glucose every few minutes in the fluid under the skin, not in blood.² GMI takes the average of those readings and runs it through one formula: GMI (%) = 3.31 + 0.02392 × average glucose in mg/dL.³ Researchers drew that line by matching many people’s average sensor glucose against their lab A1C taken at the same time.⁴ So GMI shows what A1C a typical person with your average glucose would have, not your personal A1C. A GMI of 7.0% doesn’t mean your next lab A1C will be 7.0%; it works better as a trend gauge you can check between lab visits. The number was renamed GMI, replacing “estimated A1C,” so people wouldn’t expect it to equal their lab result.⁵

You’ll usually find GMI on the summary or AGP report in your app or clinic printout, next to average glucose and time in range. It recalculates as new sensor days come in, so it shifts from week to week. The report shows the period it covers, often the last 14 days. If you use mmol/L, it may show GMI in mmol/mol instead of percent.

A sensor average can also differ from a blood average for simple reasons. Fluid under the skin lags behind blood glucose by several minutes, so the sensor trails fast rises and drops, though the lag mostly evens out over two weeks. Sensors have their own measurement error, and brands read a little higher or lower than each other, so GMI from one device may differ slightly from another’s.

These values show how average glucose converts:

To read your own number, find your average glucose on the same report as GMI and check it against these values. If they don’t line up with the formula, look at which report period each one uses. The conversion also works backward: if you have a GMI goal, you can work out the average glucose that matches it.

How Often GMI and Lab A1C Disagree

In the data used to create GMI, the two numbers matched exactly only about 1 time in 5. They differed by 0.3 points or more about half the time, and by 0.5 or more about 28% of the time.¹⁰ In an everyday clinic setting the gaps were larger: half of patients had gaps of 0.5 points or more, and about 1 in 5 had gaps of 1 point or more.¹¹ In children and teens with type 1 diabetes, the gap was meaningful in more than a third.¹²

A mismatch doesn’t mean the sensor or the lab is broken. Your clinician can compare the two over several visits to see whether your gap is stable. A stable gap is simply your normal, while a sudden new gap is worth asking about. A half-point difference can move you across a 7% target line, so neither number should drive a dose change alone.

Neither number is reliably higher. In one small clinic group, the lab A1C was higher in about half of patients and GMI was higher in most of the rest.¹³ There is a pattern across groups, though: GMI tends to read above lab A1C when A1C is low and below it when A1C is high.¹⁴ Agreement is usually best in the middle, roughly A1C 6 to 8%, where nearly all GMI values fell within 1 point.¹⁵ That’s a tendency, not a rule for any one person, so your own pairs of readings tell you more. Even in the middle range, a gap of a few tenths is ordinary.

The two numbers also look at different time windows. Lab A1C reflects roughly two to three months and leans toward the more recent weeks, while GMI covers only the days your sensor was worn. If your routine changed recently, say with a new medicine, a diet change, or an illness, the two are describing different stretches of your life and can disagree for that reason alone.

Why Your GMI and A1C Don’t Match

Lab A1C depends on how long your red blood cells live and how readily sugar attaches to them, not just on glucose. An estimate that also accounted for each person’s red cell lifespan matched lab A1C far better, within about 0.1 points on average versus about 0.5 for GMI.¹⁶ If your red cells live longer than average, they collect more sugar over time and your A1C reads higher than your glucose would suggest. If they turn over quickly, they have less time to collect sugar and A1C reads lower.

Background plays a role too. GMI tends to run higher than A1C in white people and lower in Black and Chinese Asian people.¹⁷ Black youth with type 1 diabetes had higher A1C at the same average glucose than white youth.¹⁸

Health conditions can shift A1C as well. In people with diabetes, kidney disease not on dialysis, and anemia, lab A1C ran about 0.75 points above GMI on average.¹⁹ About 4 in 10 had an A1C of 7% or higher even though their GMI was on target.²⁰ Anemia and kidney disease can change how many red cells you have and how long they last, which moves lab A1C away from true average glucose. Different types of anemia push A1C in different directions, so the effect isn’t the same for everyone. That’s why a clinician may lean on sensor data more when these conditions are present.

Recent blood loss or a transfusion matters too. Heavy bleeding or a transfusion brings in younger or donor red cells, which can make A1C read lower than your glucose really has been. The effect fades over a few months as your cell population settles. Tell your clinician about any such event before they interpret a mismatch.

Sensor-side problems can also pull GMI off. Lying on the sensor during sleep can squeeze it and read falsely low, which over time can lower GMI.²¹ A loose sensor rubbing the skin, or inflammation at the insertion site, can also pull readings down.²² On sensors needing fingerstick calibration, calibrating right after a meal or insulin dose can push GMI too high or too low.²³ And big glucose swings weaken how well GMI tracks A1C.²⁴

Which Number to Trust

Lab A1C minus GMI gives your personal gap.²⁵ That gap tends to stay stable for a person over time, regardless of glucose level or treatment.²⁶ Four or more paired readings can help sort someone into a high-gap or low-gap group.²⁷ This matters because a person with a high gap may be treated more aggressively since A1C looks high, and people with high gaps had more low-glucose episodes, including at night.²⁸ If your A1C runs well above your GMI, ask before intensifying treatment based on A1C alone.

To find your own gap, write down the GMI from the same period each time you get a lab A1C, ideally the 14 days before the draw. Subtract GMI from A1C and keep the results. Bring your CGM report to each lab A1C visit and ask your clinician to record both numbers side by side. After several visits you’ll see whether your gap is consistently positive, negative, or near zero, and a consistent gap helps you and your clinician read one number in light of the other.

A1C remains the lab standard because much of the research linking diabetes complications to glucose control was built on it, so treatment targets are traditionally framed around it. GMI is derived from sensor data and isn’t a replacement for that lab test.

What GMI adds is speed. It updates after a couple of weeks, so you can see whether a medication or habit change is working long before the next lab draw. It also comes from your own glucose readings rather than from how your red cells behave, which makes it especially useful when something is likely to distort A1C.

Neither number shows lows. Both are averages, so frequent lows can hide behind a good number. Check time below range on your report separately, and report repeated lows even if GMI looks fine.

How Many Days of Sensor Data GMI Needs

GMI from 7 days of data differed very little from GMI using 14 days for most people, but 10 to 14 days is still preferred.²⁹ In youth with type 1 diabetes, results reliably landed on the same side of target thresholds only with 10 days or more.³⁰ Glucose varies by day with meals, activity, stress, and sleep, so a longer window gives a steadier average. A GMI after a few days of a new sensor, or with big gaps in wear, deserves less weight.

To keep your GMI reliable, wear the sensor continuously and replace it promptly when it ends. Check the report’s percentage of time the sensor was active. An unusual week, such as illness, surgery, or travel, can skew the number, and critical illness or swelling can throw readings off temporarily. If you’ve worn the sensor less than a week, treat the GMI as a rough hint rather than a figure to compare with your lab result.

GMI and Time in Range

Time in range (TIR) is the share of the day your glucose stays between 70 and 180 mg/dL.³¹ In thousands of pump users, average time in range above 70% went with GMIs of about 6.9% to 7.3%, meeting common A1C targets.³² A tighter time in range of 70 to 140 mg/dL predicted reaching a GMI under 7% or 6.5% better than standard time in range.³³ But at the same GMI, time in range differs depending on how much glucose swings.³⁴

GMI only looks at the average, so someone with steady glucose and someone swinging between highs and lows can get the same number. Time in range reveals this, which GMI can’t.

Read together, a target GMI with low time in range suggests large swings. Rising time in range usually brings GMI down over weeks, and time in range responds faster to changes than A1C does.

GMI in Pregnancy and Childhood

In pregnancy with type 1 diabetes, GMI was less accurate, especially in the first and third trimesters.³⁵ It predicted pregnancy outcomes no better than average glucose and pregnancy time in range of 63 to 140 mg/dL.³⁶ GMI did look similar to A1C in women with type 1 and gestational diabetes in one comparison.³⁷ Even so, in pregnancy, time in range and time above 140 mg/dL are the more useful numbers to follow.

Pregnancy changes A1C itself, since red blood cell turnover shifts. Targets are also tighter, so small errors matter more. Ask your team which number they want you to watch.

Pediatric A1C targets are often around 7.5% rather than 7%. Parents do best sharing both numbers with the care team rather than choosing one.