Your CGM app has been telling you 6.4 for months. You did the work: the walks, the swaps, the sensor on your arm through every shower and sleeve. Then the lab result posts to the portal and says 7.1, and the floor drops. Did the sensor lie to you? Did the lab botch the draw? Have you been celebrating a number that was never real? Take a breath: this mismatch is one of the most common experiences in modern diabetes care, it has known causes, and most of the time both numbers are correct. They are answering different questions.
What do A1C and GMI actually measure?
A1C is a blood test measuring the percentage of your hemoglobin with glucose stuck to it. Red blood cells live about three months, so the test reflects your average glucose over that window, weighted toward the most recent weeks. GMI (glucose management indicator) never touches your blood. It is a formula, GMI = 3.31 + 0.02392 x mean glucose, that converts your average sensor reading (typically the last 14 or more days) into an A1C-shaped number, built so CGM users could relate sensor data to the lab test they already knew. So one number is chemistry happening on your red blood cells, and the other is arithmetic on your sensor data. Same units, different questions. (Your lab report may also show an eAG, which runs the translation the other direction; the A1C guide covers that conversion.) One more thing worth knowing: the GMI formula was built on group averages, and you are one person.
Why your A1C and CGM average diverge
- Red blood cell lifespan, the big one. The formula assumes an average lifespan. If your red cells live longer than average, glucose has more time to accumulate on them and your lab A1C reads higher than your CGM suggests. Shorter-lived cells read lower. This is stable individual biology; some people simply glycate high or low.
- Anemia and iron deficiency. Iron-deficiency anemia can push A1C falsely high; conditions or treatments that speed up red cell turnover pull it low.
- Kidney disease. Can distort A1C in either direction, through both red cell effects and treatment effects like erythropoietin or dialysis.
- Pregnancy. Red cell turnover speeds up, so A1C tends to read lower than average glucose implies. This is one reason pregnancy care leans on CGM and glucose targets over A1C alone.
- Recent illness, blood loss, or transfusion. Anything that replaces or destroys red cells resets the slow accumulation A1C depends on.
- Different time windows. GMI usually summarizes two weeks; A1C summarizes three months. If your control recently changed, in either direction, the two should disagree for a while.
- CGM wear gaps and sensor bias. A GMI built from 60% wear time, missing the nights or the chaotic weeks, summarizes a filtered version of your life. Sensors also read interstitial fluid with a lag and can run slightly high or low for an individual arm.
How big a gap is normal, and which number should you trust?
A difference of a few tenths of a point is ordinary and requires no explanation. In the research behind the GMI, plenty of people showed gaps larger than that; a gap of around half a point or more, consistently, in the same direction, is the threshold worth an explicit conversation, because it likely means your personal biology offsets the lab number and your targets should be read with that offset in mind. What a persistent gap is not: proof you failed, or proof your equipment did.
As for trust, give each number its actual job. A1C is the validated anchor for diagnosis and for the decades of research linking average glucose to long-term complications; it is the number the ADA Standards of Care frame most targets around. The CGM shows the shape of your days: the overnight lows, the post-meal spikes, the same meal landing differently on different days. An average can hide a day that swings from 55 to 250 behind the same number as a steady 140, and the CGM is the only one of the two that can tell those apart. Long-term risk: lean A1C. Daily safety and patterns: lean the sensor.
Time in range: the third lens that settles arguments
When the two averages argue, time in range (TIR) often answers the question both were circling: how much of your day is actually spent where you want to be. TIR is the percentage of readings inside a target band, commonly 70 to 180 mg/dL, with many care teams aiming for above 70% in range for adults with diabetes, individualized for age, pregnancy, and hypoglycemia risk. It also splits out time below range, which neither A1C nor GMI can see at all. If your A1C disappointed you but your TIR is strong and your lows are rare, that is a genuinely reassuring picture, and it is exactly the kind of nuance the mismatch conversation should surface. Your target band is set with your care team; the percentages are theirs to interpret.
What to bring to the appointment when they disagree
- Both numbers, dated. The lab A1C and the GMI from the same stretch, plus the previous pair if you have them, so the gap's consistency is visible.
- Your sensor wear percentage. In the CGM app's report. A GMI from 95% wear means something different than one from 60%.
- The full CGM summary, ideally 90 days. Time in range, time below range, and the overnight pattern, so the conversation covers the shape and the average together.
- Anything that touches red blood cells. Anemia, iron status, kidney disease, pregnancy, recent illness, transfusion, or a hemoglobin variant in your family. If a CBC came back recently, bring it.
- One direct question: "given my gap, which number should my targets be based on?" Some clinicians will anchor on the CGM data, some on an adjusted read of the A1C. Either is legitimate; what matters is that your plan picks a lane on purpose.
How Kite handles this
Text Kite both numbers ("lab A1C 7.1, GMI says 6.4") and it lines them up against your history, explains the likely reasons for your gap in plain language, and flags the conditions worth mentioning, like anemia or kidney disease, that bend the lab number. Before the visit, it turns your readings and questions into a one-page summary so the mismatch becomes a five-minute conversation instead of a spiral. Kite never grades your numbers and never touches your treatment; it makes sure nothing gets lost before the person who can. Text Kite to start.