How do I convert an HbA1c to an estimated average glucose?

The clinician who ordered the panel was not concerned; I would still like to understand it.

I want the working, not the result — I need to be able to redo it with different numbers.

I care about the precision as well as the value — I want to know how many figures are real.

Where is my error, and what is the correct working?

The honest answer is that the number will move, that it will move late, and that testing it early produces a figure that will be misread as a plateau.

The estimated-average-glucose relationship in common use is eAG in milligrams per decilitre equal to 28.7 times HbA1c in per cent minus 46.7. At 7.0 per cent that gives 154 mg/dL, which is 8.6 mmol/L.

Fructosamine and glycated albumin cover roughly two to three weeks and are the appropriate tests when the three-month window is either too long or unreliable.

The fifty-twenty-five-twenty-five weighting across the preceding three months is a standard result from red-cell kinetics and is the basis of the usual clinical advice to retest at three months.

Check for anything affecting red cells before concluding the number is wrong about you.

Answer first: HbA1c is a weighted average over the lifespan of circulating red cells, and roughly half of the value reflects the most recent thirty days. That weighting is the answer to most questions about lag.

Anything that shortens red-cell lifespan lowers HbA1c independently of glucose. Haemolysis, recent blood loss, iron-deficiency correction and pregnancy all do this, and the result is a falsely reassuring number.

Anything that lengthens red-cell survival raises it. Iron-deficiency anaemia before treatment, splenectomy and vitamin B12 deficiency are the usual examples, and they push in the opposite direction.

The eAG conversion in common use comes from the A1c-Derived Average Glucose study, which regressed continuous-monitoring data against measured HbA1c across a large cohort.

Half the value is the last month. That single fact answers most questions in this tag.

It helps to be literal here: a single HbA1c is a summary of about three months of glucose. Repeating it after four weeks measures mostly the same three months again.

A continuous glucose monitor gives glucose management indicator and time-in-range within a fortnight, which is why it is the better instrument for judging whether something is working at four weeks.

Point-of-care and laboratory HbA1c assays are standardised to the same reference, but between-assay variation of a couple of tenths of a point is normal and is smaller than most changes people are trying to detect.

Averages conceal excursions. A comfortable HbA1c is compatible with substantial postprandial spikes and with hypoglycaemia, and both are invisible in the single number.

Convert to average glucose before comparing against meter data, or you are comparing two different units.

edited 2 Mar 2026 by kwn_analytical — removed a claim I could not source

The lag is not a property of the drug. It is a property of the assay, and it would be the same lag for any intervention that changed mean glucose by the same amount.

Red cells live roughly a hundred and twenty days, but the glycation is cumulative, so the contribution is not uniform. The commonly used weighting puts about fifty per cent of the value in the preceding thirty days, about twenty-five per cent in days thirty-one to sixty, and the remainder in the two months before that.

Retest at three months. Testing at four weeks tells you mostly what the previous two months were doing.

edited 16 Jun 2026 by Dr_Lena_Ostrowska — fixed an arithmetic slip in the third paragraph