What Your HbA1c Number Does and Does Not Tell You

Quick answer

HbA1c reflects average glucose exposure over roughly the previous ten to twelve weeks and diagnoses diabetes at 48 mmol/mol or above, but it is not appropriate in pregnancy or acute illness and is unreliable in anaemia, haemoglobin variants, chronic kidney disease and after a recent transfusion. Do not assume the number describes your day-to-day glucose pattern, and tell your clinician about any of these before medication is changed on the basis of a result.

HbA1c has become shorthand for metabolic health, and it is a genuinely useful test: standardised, reproducible, and strongly linked to long-term complication risk. It is also routinely over-read. A single number is treated as a complete description of how someone's glucose behaves, which it is not.

There are two separate issues. The first is that HbA1c is sometimes unreliable, because the measurement depends on red blood cells and anything that alters those cells alters the result. The second is that even when the measurement is accurate, two people with identical HbA1c values can have very different day-to-day glucose patterns.

Both are worth understanding before acting on a result.

What HbA1c measures

Glucose in the blood attaches slowly and irreversibly to haemoglobin inside red blood cells. HbA1c is the proportion of haemoglobin that has been glycated in this way, reported in the UK in mmol/mol.

Because red cells survive for around 120 days and are replaced continuously, the result reflects average glucose exposure over roughly the preceding ten to twelve weeks, weighted towards the more recent weeks. That weighting is why a few disciplined days before a blood test change very little, and why a genuine improvement takes a couple of months to show fully.

The UK diagnostic threshold

Diabetes is diagnosed at an HbA1c of 48 mmol/mol or above, equivalent to 6.5 per cent. Where there are no symptoms, a diagnosis is not normally made on one result alone.

A value below 48 mmol/mol does not exclude diabetes. Fasting plasma glucose of 7.0 mmol/l or more, or a two-hour value of 11.1 mmol/l or more after a 75 g glucose load, also makes the diagnosis, and glucose-based testing will identify some people whom HbA1c misses.

NICE advises measuring HbA1c every three to six months in adults with type 2 diabetes until it is stable on unchanging treatment, then every six months.

When HbA1c should not be used to diagnose diabetes

UK guidance is explicit that HbA1c is not appropriate for diagnosis in several situations. These include:

Children and young people, and anyone of any age in whom type 1 diabetes is suspected
Symptoms of diabetes present for less than two months
Acute illness, particularly where hospital admission is needed
Acute pancreatic damage, including pancreatic surgery
Pregnancy
Any genetic, haematological or illness-related factor that affects haemoglobin or its measurement

That last category covers a good deal of ground, and it is where results are most often misread.

Conditions that make the number unreliable

Anything that shortens red cell survival reduces the time available for glycation and tends to lower HbA1c, which can make glucose control look better than it is. Anything that lengthens average red cell age tends to raise it.

Factors reported to lower HbA1c, or to make it unreliable, include haemolytic anaemia, increased red cell fragility, recent blood transfusion, reticulocytosis, chronic liver disease, treatment with erythropoietin, and iron or vitamin B12 replacement in someone who was deficient. Factors reported to raise it include untreated iron deficiency, untreated vitamin B12 deficiency, splenectomy and chronic kidney disease. Haemoglobin variants, raised fetal haemoglobin and methaemoglobin can shift the result in either direction, and may also interfere with the assay itself depending on the method used.

Where HbA1c monitoring is invalid because of disturbed red cell turnover or an abnormal haemoglobin type, NICE recommends estimating trends by other means: quality-controlled plasma glucose profiles, total glycated haemoglobin where haemoglobins are abnormal, or fructosamine.

The practical point is that an unexpected result, or one that does not fit the clinical picture, should prompt a question about the red cells rather than an immediate change of treatment.

Why two people with the same HbA1c can differ

Even a technically accurate HbA1c is an average, and averages conceal distribution.

A study published in Diabetes Care in 2017 pooled continuous glucose monitoring data from 387 participants across three trials. For an HbA1c of 8 per cent, roughly 64 mmol/mol, the corresponding mean glucose spanned approximately 8.6 to 12.1 mmol/l. The authors noted that this range overlapped substantially with the ranges for 7 and 9 per cent, so the same HbA1c could reflect good, fair or poor average glucose depending on the individual.

Part of this is biological variation in how readily haemoglobin glycates. Part of it is that an average says nothing about shape. One person may sit steadily just above target all day. Another may swing between hypoglycaemia overnight and high post-meal peaks and arrive at the same average. Those two situations carry different symptoms, different risks and different treatment implications, and HbA1c cannot distinguish them.

How to use the number well

HbA1c remains the right test for diagnosis in most adults and the right test for tracking change over months. It is a poor tool for answering "what is happening to my glucose today", and it was never designed to do that.

If your result does not match how you feel, or if you have anaemia, kidney disease, a haemoglobin variant, a recent transfusion or a pregnancy, say so before any treatment decision is made. Where the picture remains unclear, glucose profiles or continuous monitoring can show the pattern that the average hides. Do not adjust glucose-lowering medication on the basis of a single HbA1c without clinical review.