HbA1c: what a three-month average can and cannot show

HbA1c has an unusual property among blood tests: it does not depend on when you last ate. That single feature is why it became central to diabetes care.

It works by measuring how much glucose has attached itself to haemoglobin inside red blood cells. Because that attachment is slow and essentially permanent for the life of the cell, the amount accumulated reflects average glucose over the preceding period.

Why three months

Red blood cells live around 120 days on average, and the population in circulation at any moment spans that whole range. The measurement therefore reflects glucose over roughly the preceding two to three months.

The weighting is not even. Recent weeks contribute more than distant ones, because more of the currently circulating cells were exposed during them. A change made a month ago is already partly visible; one made a week ago is barely reflected.

This is why repeating HbA1c a few weeks after a change is usually unhelpful, and why the conventional interval is around three months.

The first thing an average hides

An average says nothing about variability. Two people with identical HbA1c can have very different daily patterns — one steady, one swinging between high and low readings that cancel out.

This matters clinically, and it is one reason continuous glucose monitoring has become more common alongside HbA1c rather than replacing it. The two measure different aspects of the same thing.

The second thing: red cells vary too

The measurement assumes red cells live a typical length of time. Anything that shortens that lifespan reduces the accumulated glucose and lowers HbA1c independently of blood sugar.

Conditions that shorten red cell survival, recent significant blood loss, and recent transfusion all affect the result. So does pregnancy, where red cell turnover changes.

In the other direction, conditions that lengthen red cell survival can raise HbA1c without any change in glucose.

This is why an HbA1c that does not match the glucose readings a person is seeing is worth discussing rather than assuming one of them is wrong.

Haemoglobin variants

Some inherited haemoglobin variants interfere with certain HbA1c methods, producing results that are inaccurate in ways the report does not show.

Laboratories know which variants affect their particular method, and alternative approaches exist. Where a result is persistently out of keeping with other measures of glucose, this is one of the explanations a clinician will consider.

Why it is used for diagnosis as well as monitoring

For many years HbA1c was used only to monitor people already known to have diabetes, and diagnosis rested on glucose testing. That changed as the measurement became better standardised between laboratories.

Its practical advantages for diagnosis are considerable: no fasting, no timed drinks, no requirement to attend early in the morning, and a single sample that can be taken alongside anything else. Glucose testing requires conditions that are inconvenient enough that people skip them.

Diagnostic thresholds are decision points chosen by expert consensus, not population reference ranges. They mark where the balance of evidence favours acting, which is a different kind of number from the middle 95 per cent of a population, and confirmation on a second sample is conventional before a diagnosis is made on one alone.

Where glucose testing still does more

HbA1c is slow by design, which makes it poor at answering urgent questions. Where blood sugar is changing rapidly, an average over months is precisely the wrong instrument.

It also cannot detect low blood sugar. Episodes of hypoglycaemia can be frequent and clinically important while contributing to an average that looks unremarkable, or even to one that looks better than it should.

This is why HbA1c and day-to-day glucose readings are complements rather than alternatives, and why a good HbA1c alongside symptomatic swings is a conversation worth having rather than a contradiction to resolve.

Two scales

HbA1c was historically reported as a percentage and is now widely reported in millimoles per mole. Many reports print both, and a great deal of older material online uses only the percentage.

The two are not proportional, so converting by eye does not work. Comparing a current result against an older one recorded in the other unit requires an actual conversion rather than an estimate.