Glucose in the blood attaches itself to haemoglobin, the protein that carries oxygen in red blood cells. Once attached it stays attached for the life of that cell, about three months. Measure what fraction of your haemoglobin is glycated and you get HbA1c — a weighted average of your glucose over roughly the last eight to twelve weeks, with the most recent month counting for about half.
What the number means
| HbA1c | Estimated average glucose | |
|---|---|---|
| 5.0% | 97 mg/dL · 5.4 mmol/L | Normal |
| 5.7% | 117 mg/dL · 6.5 mmol/L | Prediabetes threshold |
| 6.5% | 140 mg/dL · 7.8 mmol/L | Diabetes threshold |
| 7.0% | 154 mg/dL · 8.6 mmol/L | Common adult target |
| 8.0% | 183 mg/dL · 10.2 mmol/L | |
| 9.0% | 212 mg/dL · 11.8 mmol/L | |
| 10.0% | 240 mg/dL · 13.4 mmol/L |
The conversion is eAG (mg/dL) = 28.7 × A1c − 46.7, from the ADAG study. Sukkari's calculator does it in both directions.
What target to aim for
Below 7% is the general adult target, but "general" is doing real work in that sentence. A younger person with a long life ahead and no history of severe lows may be aimed at 6.5%. An older person, someone with heart disease, or someone who does not reliably feel their lows may be safer at 7.5% or 8%. A target that is too tight is not automatically better — pushing A1c down at the cost of frequent hypoglycaemia is a bad trade, and the large trials showed exactly that.
GMI: the same idea, from your own meter
The Glucose Management Indicator estimates what your A1c is likely to be, using the average of the readings you have logged: GMI = 3.31 + 0.02392 × mean glucose in mg/dL. Its advantage is immediacy — you see the effect of a change in a couple of weeks rather than waiting for the next lab draw.
Its limitation is that it only knows the readings you gave it. If you mostly test fasting, your GMI will look better than your real A1c, because you never sampled the post-meal peaks. If you mostly test when you feel unwell, it will look worse. GMI is honest only in proportion to how representatively you sample.
When A1c and GMI disagree
A gap of 0.3% or so is normal. A larger gap is usually explained by one of:
- Sampling bias — the most common cause, described above.
- Red cell lifespan. Anything that shortens it makes A1c read low: iron-deficiency anaemia under treatment, haemolysis, recent blood loss, chronic kidney disease, pregnancy. Anything that lengthens it reads high.
- Haemoglobin variants. Certain thalassaemia and sickle traits interfere with some A1c assays. If your A1c never seems to match your daily numbers, this is worth asking about.
- Individual glycation rate. Some people simply glycate faster or slower than the population average.
A1c and your daily log are not competitors. A1c is the audited annual accounts; your log is the daily ledger that tells you which transactions caused the result. You need both.