GMI (Glucose Management Indicator) is a CGM-derived estimate of A1C calculated from mean glucose. GMI and laboratory A1C usually correlate but can differ by 0.3 to 0.5 percent for legitimate physiologic reasons. Both are useful — GMI for fast feedback, A1C for clinical decision-making.
What GMI Is
- Derived from CGM data over a defined window (typically 14 days).
- Formula: GMI (%) = 3.31 + 0.02392 × mean glucose (mg/dL).
- Published by Bergenstal et al. in Diabetes Care, 2018.
- Replaced the older term “estimated A1C” (eA1C), which was confused with laboratory A1C.
- Automatically displayed by Dexcom Clarity, FreeStyle LibreView, and other CGM platforms.
- Requires at least 70 percent sensor wear (10 of 14 days) for reliable values.
What A1C Is
- Glycated hemoglobin — a blood test measuring how much glucose has bound to hemoglobin A.
- Reflects average blood glucose over the past 2 to 3 months (the lifespan of red blood cells).
- Reported as a percentage of total hemoglobin (also reportable as mmol/mol in many countries).
- Reference standard for diabetes diagnosis (A1C ≥ 6.5%), prediabetes (5.7 to 6.4%), and complications risk assessment.
- For background on diagnostic ranges, see our guide on A1C levels.
Side-by-Side Comparison
| Feature | GMI | A1C |
|---|---|---|
| Source | CGM data | Blood test |
| Window | 14 days (typically) | 2 to 3 months |
| Frequency available | Daily or weekly | Quarterly (typically) |
| Affected by hemoglobin biology | No | Yes |
| Affected by sensor accuracy | Yes | No |
| Used for diagnosis | No | Yes |
| Used for therapy adjustments | Yes (rapid) | Yes (over time) |
| Reflects variability | No (it is a mean) | No |
| Captures hypoglycemia | No | No |
Worked Examples
| 14-Day Mean Glucose (mg/dL) | GMI Calculation | GMI (%) |
|---|---|---|
| 120 | 3.31 + 0.02392 × 120 | 6.18 |
| 140 | 3.31 + 0.02392 × 140 | 6.66 |
| 154 | 3.31 + 0.02392 × 154 | 6.99 |
| 170 | 3.31 + 0.02392 × 170 | 7.38 |
| 183 | 3.31 + 0.02392 × 183 | 7.69 |
| 200 | 3.31 + 0.02392 × 200 | 8.09 |
| 240 | 3.31 + 0.02392 × 240 | 9.05 |
Why GMI and A1C Can Differ
- Hemoglobinopathies: sickle cell trait, sickle cell disease, thalassemia, hemoglobin C — alter glycation patterns and A1C assay results.
- Anemia or iron deficiency: changes red blood cell turnover; iron deficiency tends to raise A1C, while hemolytic anemia tends to lower it.
- Recent blood transfusion: introduces non-glycated red cells, temporarily lowering A1C.
- Erythropoietin (EPO) therapy or chronic kidney disease: shortens red cell lifespan, lowering A1C.
- Pregnancy: increased red cell turnover and plasma volume lower A1C — pregnancy-specific TIR targets are used.
- Recent acute illness: alters glucose patterns transiently.
- Individual glycation rate variability: some people glycate hemoglobin faster than others at the same mean glucose (“high glycators” vs “low glycators”).
- CGM sensor bias: sensors can read consistently higher or lower than venous glucose, especially in the first 24 hours of wear.
When to Trust Which
| Clinical Question | Best Metric |
|---|---|
| Diagnosis of diabetes or prediabetes | A1C (or fasting glucose, OGTT) |
| Day-to-day therapy adjustments | GMI plus TIR |
| Complications risk benchmarking | A1C |
| Pregnancy glucose management | GMI plus TIR (pregnancy ranges) |
| Patient with hemoglobinopathy | GMI plus TIR (A1C unreliable) |
| Hospitalized or unstable patient | CGM data directly; GMI less useful |
| Annual clinic visit benchmarking | Both, side by side |
How to Use GMI Clinically
- Verify CGM wear exceeds 70 percent of the window (about 10 of 14 days).
- Note the GMI value and compare to most recent A1C.
- If GMI and A1C differ by less than 0.3 percent, treat them as consistent.
- If they differ by more than 0.5 percent, investigate causes: hemoglobinopathy, anemia, transfusion, EPO, pregnancy, sensor calibration.
- Use GMI for short-term adjustments (insulin dose, food timing, AID settings).
- Use A1C for periodic check-ins and to assess complications risk.
- Document the discrepancy direction — patients should know if their A1C consistently runs higher or lower than GMI.
Common Clinical Scenarios
- GMI 7.0%, A1C 7.5%: mild discrepancy — usually a “high glycator” or sensor reading slightly low. Track trend over several visits.
- GMI 7.0%, A1C 6.5%: potential anemia, hemolysis, EPO use, or shortened RBC lifespan.
- GMI 7.5%, A1C 8.5%: consider iron deficiency or sensor reading low; check CBC and iron panel.
- GMI matches A1C, but TIR is poor: high variability is the issue — see CV for glucose and MAGE.
- GMI dropping rapidly, A1C lagging: normal — A1C takes 2 to 3 months to catch up to behavior change.
Limitations of GMI
- Requires CGM with adequate wear.
- Depends on sensor accuracy — bias of 5 to 10 mg/dL translates to GMI bias of 0.1 to 0.2 percent.
- Formula was derived from predominantly type 1 diabetes data; type 2 generalization is reasonable but imperfect.
- Single mean glucose loses variability information — pair with TIR, TBR, and CV.
- Not currently a diagnostic criterion.
- 14-day windows may be too short during periods of rapid change.
Practical Tips
- Check GMI weekly if you wear a CGM continuously.
- Print the AGP report before each clinic visit — it shows GMI, TIR, and A1C side by side.
- If GMI runs consistently lower than A1C, ask your clinician about hemoglobinopathy screening.
- Calibrate the CGM appropriately if your platform requires fingerstick calibration.
- Replace sensors on time — accuracy can drift in the final hours of wear.
- Compare GMI and TIR together — they tell different parts of the story.
Related Reading
For deeper coverage of related CGM metrics, see Time in Range, coefficient of variation, MAGE, and Ambulatory Glucose Profile. For background on testing for prediabetes, see our detection of prediabetes hub.
The Bottom Line
GMI is a CGM-derived estimate of A1C using the formula 3.31 plus 0.02392 times mean glucose. It usually correlates with laboratory A1C but can differ by 0.3 to 0.5 percent for legitimate physiologic reasons. Use GMI for fast, granular adjustments — it updates in days, not months. Use A1C for diagnostic and complications-risk decisions. When the two disagree, look for hemoglobinopathies, anemia, transfusion, EPO use, pregnancy, or sensor calibration issues. The two metrics complement each other; neither replaces the other.
