Diabetes (well controlled)
A1C below 7% is the target for most adults with diabetes.
| A1C (%) | Avg. Glucose (mg/dL) | Avg. Glucose (mmol/L) |
|---|---|---|
| 5.0% | 97 | 5.4 |
| 5.5% | 111 | 6.2 |
| 6.0% | 126 | 7 |
| 6.5% | 140 | 7.8 |
| 7.0% | 154 | 8.6 |
| 7.5% | 169 | 9.4 |
| 8.0% | 183 | 10.2 |
| 8.5% | 197 | 10.9 |
| 9.0% | 212 | 11.8 |
| 9.5% | 226 | 12.6 |
| 10.0% | 240 | 13.4 |
GMI may differ from lab A1C due to individual variations in red blood cell lifespan and hemoglobin glycation. Formula: GMI = 3.31 + 0.02392 × mean glucose (mg/dL).
GMI, or Glucose Management Indicator, is a metric that estimates what your A1C would likely be based on your average blood glucose readings from a continuous glucose monitor (CGM). Previously called "estimated A1C" or "eA1C," the term was changed to GMI in 2018 to help differentiate it from laboratory-measured A1C.
The GMI provides a way to bridge the gap between the continuous data from CGM systems and the traditional A1C test that's been used for decades to assess long-term blood glucose control. While A1C measures the percentage of hemoglobin in your blood that has glucose attached to it over the past 2-3 months, GMI offers a more immediate estimate based on your actual glucose readings.
The GMI formula was developed by Bergenstal and colleagues in 2018, using data from 528 participants across four clinical trials with Dexcom G4 CGM sensors:
For those using mmol/L:
If your CGM shows an average glucose of 150 mg/dL over the past 14 days:
One of the most common questions about GMI is why it sometimes differs from laboratory A1C. Understanding these differences is important for proper interpretation.
Red blood cell lifespan variation: A1C measures glucose attached to hemoglobin in red blood cells. If your red blood cells live longer or shorter than average (about 120 days), your lab A1C may be higher or lower than expected.
Hemoglobin variants: Certain genetic variations in hemoglobin structure can affect how glucose binds, potentially skewing lab A1C results.
Recent glucose fluctuations: GMI reflects your glucose average over the CGM wear period (ideally 14+ days), while A1C reflects approximately 3 months of glucose exposure, with more recent weeks weighted slightly more.
Medical conditions: Conditions affecting red blood cells, such as anemia, kidney disease, or liver disease, can impact lab A1C accuracy.
CGM accuracy: While modern CGMs are highly accurate, some measurement error exists that can affect the calculated average.
Research shows that about half of people with diabetes have a meaningful difference between their GMI and lab A1C. The difference can range from -1.0% to +1.0% or even more. Neither value is necessarily "wrong"—they're measuring slightly different things and can both provide valuable information.
| A1C/GMI (%) | Classification | Target population |
|---|---|---|
| Below 5.7% | Normal | Non-diabetic |
| 5.7%-6.4% | Prediabetes | At risk for diabetes |
| 6.5%-7.0% | Well-controlled diabetes | Most adults with diabetes |
| 7.0%-8.0% | Moderate control | May be appropriate for some |
| Above 8.0% | Needs improvement | Discuss with healthcare provider |
The American Diabetes Association recommends an A1C target of below 7% for most adults with diabetes. However, targets may be individualized based on:
Some individuals may have higher targets (7.5%-8%) if hypoglycemia is a concern, while others aiming for pregnancy or certain other situations may target lower values (6%-6.5%).
The relationship between average glucose and A1C/GMI is roughly linear. Here's what different A1C levels correspond to in terms of average glucose:
| A1C (%) | Avg. Glucose (mg/dL) | Avg. Glucose (mmol/L) |
|---|---|---|
| 5.0% | 97 | 5.4 |
| 5.5% | 111 | 6.2 |
| 6.0% | 126 | 7.0 |
| 6.5% | 140 | 7.8 |
| 7.0% | 154 | 8.6 |
| 7.5% | 169 | 9.4 |
| 8.0% | 183 | 10.2 |
| 8.5% | 197 | 10.9 |
| 9.0% | 212 | 11.8 |
| 10.0% | 240 | 13.4 |
For the most reliable GMI calculation, follow these best practices:
Minimum 14 days of data: The GMI formula was validated using at least 14 days of CGM readings. Shorter periods may be less accurate.
High sensor wear time: Aim for at least 70% time wearing the sensor (about 10 days out of 14). Gaps in data can skew the average.
Calibrate as recommended: Follow your CGM manufacturer's calibration instructions if applicable.
While GMI provides a useful summary metric, modern CGM technology offers additional insights through Time in Range (TIR) metrics:
| Metric | Target | Description |
|---|---|---|
| Time in Range (70-180 mg/dL) | >70% | Hours per day in target zone |
| Time Below Range (<70 mg/dL) | <4% | Hypoglycemia risk |
| Time Above Range (>180 mg/dL) | <25% | Hyperglycemia exposure |
| Coefficient of Variation | <36% | Glucose variability |
Many diabetes experts now recommend using Time in Range alongside GMI/A1C for a more complete picture of glucose management. Someone with the same GMI can have very different glucose patterns—one person might have steady glucose levels while another experiences significant highs and lows that average out to the same value.
While GMI is a valuable tool, it has several limitations to keep in mind:
Not a replacement for lab A1C: GMI complements but doesn't replace periodic laboratory A1C testing, which may be required for insurance or clinical decisions.
Individual variation: The formula represents a population average. Your personal relationship between glucose and A1C may differ.
CGM accuracy dependence: GMI is only as accurate as your CGM data. Sensor errors, compression lows, or insufficient calibration can affect results.
Short-term vs. long-term: GMI based on 14 days doesn't capture the full 2-3 month picture that lab A1C reflects.
Not validated in all populations: The original GMI formula was derived primarily from adults with type 1 and type 2 diabetes. It may be less accurate for other populations.
The concept of estimating A1C from glucose data has evolved over several decades:
The transition to GMI acknowledged that while the metrics are related, they measure subtly different things and shouldn't be treated as interchangeable.