GMI vs A1C: Why Your App and Your Lab Disagree

Your app says your Glucose Management Indicator is 7.2%. Three days later the lab draws blood and reports an A1C of 8.0%. Nothing malfunctioned. The two numbers are produced by completely different mechanisms, and the distance between them tells you something worth knowing.

What A1C Actually Measures

A1C is a chemical measurement, not a calculation. Glucose in your bloodstream attaches to hemoglobin in your red blood cells through a process called glycation, and once attached it stays there for the life of the cell. A lab A1C counts what fraction of your hemoglobin carries glucose.

Because red blood cells live roughly three months, that fraction reflects your average glucose over the previous 60 to 90 days — weighted toward the most recent month, since those cells have had the most recent exposure.

What GMI Measures

The Glucose Management Indicator is a calculation, not a measurement. It takes the mean glucose from your CGM over a period and converts it into the A1C-like percentage you are used to reading:

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

That formula came from a study comparing CGM averages against measured lab A1C across a large group of people. It describes what A1C typically is for a given average glucose — a population relationship, applied to you.

Which is exactly where the disagreement comes from.

Why the Two Numbers Diverge

Three separate things pull them apart.

They cover different windows. GMI reflects whatever period you selected — often 14 or 30 days. A1C reflects roughly 90. If the last two weeks went well after a rough couple of months, GMI will look better than your lab result, and both will be accurate about their own timeframe.

Red blood cells vary between people. The 90-day lifespan is an average. If your cells live longer than typical, they accumulate more glucose and your A1C reads higher than your average glucose suggests. Shorter-lived cells produce the opposite. Anemia, kidney disease, pregnancy, recent blood loss and several hemoglobin variants all shift this, sometimes substantially.

Glycation rates differ between people. Some people's hemoglobin simply binds glucose more readily than others at the same exposure. Researchers describe this as being a high or low glycator. It is a stable personal trait, and it means the population formula will consistently over- or under-estimate for you specifically.

The Estimate Your App Shows May Not Be GMI

Here is a subtlety that catches people out, and it is visible inside SweetLife itself: there are two different formulas in common use, and they do not agree.

The older one, estimated A1C or eAG, comes from the ADAG study and converts average glucose like this: A1C = (average + 46.7) ÷ 28.7. GMI uses the newer CGM-derived formula above. Run the same average through both and you get different answers:

• Average 120 mg/dL — eA1C 5.8%, GMI 6.2%
• Average 140 mg/dL — eA1C 6.5%, GMI 6.7%
• Average 154 mg/dL — both give 7.0%
• Average 200 mg/dL — eA1C 8.6%, GMI 8.1%
• Average 260 mg/dL — eA1C 10.7%, GMI 9.5%

The two agree almost exactly at an average of about 154 mg/dL and diverge in both directions from there — by more than a full percentage point at the high end. So if you see two different A1C-ish numbers in the same app, or your app disagrees with a different app, this is frequently the reason. They are answering the same question with different equations.

Which Should You Trust?

For clinical decisions, the lab A1C is the measured value and the reference standard. Your care team will treat it that way.

For day-to-day management, GMI is more useful, for a simple reason: it updates continuously. A1C arrives every three months, long after the behaviour that produced it. GMI tells you where the current fortnight is heading while you can still influence it.

The most valuable thing, though, is neither number in isolation — it is the gap between them. If your GMI consistently runs 0.5% below your lab A1C across several draws, that gap is a stable personal fact. Once you and your endocrinologist know it, GMI becomes a reliable forecast of your next lab result rather than a number that mysteriously disagrees with it.

What GMI Cannot Tell You

GMI is built from an average, which means it inherits every blind spot an average has. Steady glucose and violent swings can produce identical GMIs, because the highs and lows cancel out in the arithmetic.

This is why GMI should be read next to Time in Range and coefficient of variation, never on its own. A GMI of 7.0% with a CV of 28% describes a genuinely stable fortnight. The same 7.0% with a CV of 44% describes two weeks of highs and lows averaging out to a respectable-looking number.

One further caveat: GMI is only as good as the data behind it. If your CGM had gaps, or you wore it inconsistently, the average is drawn from a partial picture. Consensus guidance suggests at least 14 days of data with roughly 70% sensor coverage before treating GMI as reliable.

How SweetLife Shows It

Insights displays GMI alongside Time in Range, standard deviation and average glucose for whatever period you choose, so the estimate always appears in the company of the numbers that qualify it. The reading count is shown too, which is what tells you whether there is enough data behind the figure to take it seriously.

When you generate a doctor report, all of it goes in together — which gives your endocrinologist what they need to work out your personal gap between GMI and lab A1C, and to interpret the next three months accordingly.

The Short Version

A1C is measured from your blood and covers about 90 days. GMI is calculated from your CGM average and covers whatever window you pick. They disagree because of timeframe, red blood cell lifespan, and personal glycation rate — and because two different conversion formulas are in circulation.

Track the gap between your GMI and your lab results over a few draws. Once you know its size and direction, the number on your phone becomes genuinely predictive.