What HbA1c actually measures — and when it misleads

Glucose in the bloodstream attaches itself to haemoglobin inside red blood cells, slowly and without any help from an enzyme. The higher the glucose, the more of it sticks. Red cells live about three to four months, so the proportion of haemoglobin carrying glucose is a running average of what the blood has been exposed to over that period. That proportion is HbA1c.

Why it is used

No fasting. No timing. No standing in a queue at 7 a.m. It is well standardised internationally through the NGSP programme, and it varies far less from day to day than a single glucose reading. WHO accepted it as a diagnostic test in 2011, with an HbA1c of 6.5% or above as the cut-off, while being explicit that a value below 6.5% does not rule diabetes out.

Two units, one test

Older reports give a percentage (NGSP units). Newer and international reports may give mmol/mol (IFCC units). They are the same measurement: mmol/mol = (percentage − 2.15) × 10.929. An HbA1c of 6.5% is 48 mmol/mol. The chart converts both, alongside estimated average glucose.

What it is not

It is not a glucose reading, and estimated average glucose is not a prediction of tomorrow morning's number. Two people with an identical HbA1c can have very different daily patterns — one steady, one swinging between highs and lows — and the average cannot tell them apart. It also says nothing about the last two weeks specifically; recent glucose is weighted more heavily than three-month-old glucose, but the number is still an average.

When HbA1c is the wrong test

Because the measurement depends on red blood cells, anything that changes how long red cells survive changes the result without any change in glucose. WHO's 2011 report is direct about this. Situations in which HbA1c may not be reliable include:

• Anaemia, particularly iron-deficiency anaemia — which is very common in India, and especially among Indian women.
• Inherited haemoglobin variants and thalassaemia traits, which are common in several Indian populations and can interfere with some assay methods.
• Recent blood transfusion, recent significant blood loss, or haemolysis.
• Advanced kidney disease and some liver disease.
• Pregnancy, in which red-cell turnover changes — which is one reason pregnancy is assessed with glucose testing and its own separate criteria.

None of that means an HbA1c result should be dismissed. It means the interpretation belongs to a doctor who knows whether any of it applies to you, and who can order a glucose-based test instead when it does.

Reading a result

Against the ADA criteria, below 5.7% is outside the prediabetes range, 5.7–6.4% is the prediabetes range, and 6.5% or above is the diabetes range. Those are the published bands. What follows from a result — whether it is repeated, what else is tested, what happens next — is a clinical decision, and this site does not make it.