Patient Education
Understanding HbA1c
By Miracle Olajuyigbe 7 min read
What the test is actually measuring
Glucose in your bloodstream sticks to haemoglobin, the protein inside red blood cells that carries oxygen. The reaction is slow, it is not reversible, and the amount that ends up stuck is proportional to how much glucose has been in circulation. HbA1c measures the percentage of your haemoglobin carrying that sugar coating.
The reason it covers roughly three months comes down to the lifespan of a red blood cell, which is about 120 days. Your bloodstream holds cells of every age, from new ones released this morning to old ones about to be cleared. The glucose stuck to them is a running record over that whole population, so the test captures a period rather than a moment.
One detail is often left out and it changes how you read your result. The average is weighted toward the recent past. The last month contributes something like half of the value, the month before that around a quarter, and the earliest month the remainder. Improvements you made three weeks ago do show up. A rough fortnight before the blood draw shows up too, more than you would like.
That is the practical case for the test. You can eat carefully for two days before a fasting glucose and get a flattering number. You cannot do that to an A1c.
The numbers
For diagnosis in non-pregnant adults, most guidelines use the same thresholds:
Result Interpretation Below 5.7% (below 39 mmol/mol) Normal 5.7% to 6.4% (39 to 46 mmol/mol) Prediabetes 6.5% or above (48 mmol/mol or above) Diabetes You may see two units. The percentage is the older NGSP figure used in the United States and much of the world. The mmol/mol figure is the IFCC standard used across Europe and increasingly elsewhere. Same test, different scale.
For people already diagnosed, a common target is below 7%, though that should be individualised and often is not. Tighter targets around 6.5% may suit someone young and recently diagnosed with no history of hypoglycaemia. Looser targets of 8% or above suit someone elderly or frail, or with hypoglycaemia unawareness, where the harm from a low outweighs the long-term benefit of a lower average. A target of 6.5% in an eighty-year-old on a sulfonylurea living alone is not good control. It is a fall waiting to happen.
What a single percentage point represents. A1c converts to an estimated average glucose. Each 1% change corresponds to roughly 29 mg/dL, or about 1.6 mmol/L, in average glucose. Going from 9% to 8% means your average blood sugar dropped from around 212 to around 183 mg/dL. In outcome terms, the UKPDS analysis found each 1 percentage point reduction in A1c associated with roughly 37% fewer microvascular complications, the eye, kidney, and nerve damage, and about 14% fewer heart attacks. Those are observational associations rather than a promise. Still, when a clinician says that moving from 10% to 9% is worth doing even though you are nowhere near target, that is what they are drawing on. The move you can make is worth more than the perfect number you cannot.
What A1c is good for
It is a strong test, which is worth stating before pulling it apart.
No fasting is required, so it can be done at any appointment. It is far less affected by acute illness, stress, or yesterday’s dinner than a single glucose reading, and the sample is more forgiving than fasting plasma glucose, which degrades if the tube sits around. That last point is a real advantage in laboratories without reliable cold chain. It is standardised internationally, and it predicts long-term complications well enough to have anchored diabetes care for decades.
For most people, most of the time, it does what it claims.
Where A1c can mislead
Here is the layer that patient-facing articles almost always leave out.
A1c is an indirect measurement. It infers your glucose from what has happened to your red blood cells. If anything alters the number of red cells, how long they survive, or the structure of the haemoglobin inside them, the inference breaks, and the result can be wrong in either direction while looking perfectly normal on the page.
Anything that shortens red cell survival lowers A1c. Cells cleared early have had less time to accumulate glucose, so the test reads lower than your true average. This happens with haemolytic anaemia, recent significant blood loss, a recent transfusion, splenomegaly, and in advanced kidney disease, particularly with erythropoietin treatment, which floods the circulation with young cells. Late pregnancy does the same thing through increased red cell turnover and dilution, which is one reason A1c is not used to diagnose gestational diabetes.
Anything that lengthens red cell survival raises it. Untreated iron deficiency anaemia is the common one, and it is very common indeed. Cells live longer, accumulate more glucose, and the A1c drifts up by a few tenths of a percentage point with no change in your actual glucose. Vitamin B12 and folate deficiency can do the same. This matters clinically: treat the iron deficiency and the A1c may fall on its own, which patients understandably find confusing when nothing about their diet changed.
Haemoglobin variants are the most under-discussed problem, and the most relevant if you have African, Mediterranean, Middle Eastern, South or Southeast Asian ancestry.
Sickle cell trait, haemoglobin C trait, haemoglobin E, and beta thalassaemia trait are all common in these populations. Sickle cell trait alone affects a large share of West Africans and a meaningful proportion of African Americans. Most people carrying a trait are healthy and many do not know they have it. Two separate problems follow. First, some laboratory methods misread variant haemoglobin, producing results that are too high or too low depending on the assay. Modern methods handle most variants correctly, but not all, and not everywhere. Second, in sickle cell disease itself, red cell survival really is shortened, so the A1c is biologically lower than your true glucose average regardless of how good the assay is.
The consequence is not theoretical. It means a patient can carry an A1c in the normal or prediabetic range while running glucose levels that are causing damage. It also means someone can be told they have diabetes on the strength of a falsely raised result. If you have a known haemoglobin variant, or ancestry where one is common, it is entirely reasonable to ask your clinician two questions: whether the laboratory’s method is affected by your variant, and whether a second measure should be used alongside it.
Also worth knowing: chronic liver disease, high-dose aspirin, some HIV antiretrovirals, and severe hypertriglyceridaemia can all shift the result, and age nudges A1c up slightly on its own.
What your doctor can use instead. None of these are perfect either, which is why the choice depends on the situation.
Fasting plasma glucose and the oral glucose tolerance test measure glucose directly and bypass haemoglobin entirely. The OGTT remains the reference standard and is the correct test in pregnancy.
Fructosamine and glycated albumin reflect two to three weeks rather than three months and are unaffected by red cell survival, which makes them useful in haemolytic conditions, advanced kidney disease, and pregnancy.
Continuous glucose monitoring produces a glucose management indicator derived from directly measured glucose over 14 days or more. Where available, it is the cleanest answer to the variant problem, since haemoglobin never enters the calculation.
What to do with an abnormal result
In the prediabetes range (5.7% to 6.4%): a signal, not a diagnosis, and the most actionable result on the scale. Diabetes Prevention Program data showed that structured lifestyle change, around 7% weight loss and roughly 150 minutes of activity a week, cut progression to type 2 diabetes by 58% over about three years, outperforming metformin. Have it repeated in three to six months rather than treating it as a verdict.
At 6.5% or above: guidelines call for confirmation, either a repeat A1c or a second test, unless you have clear symptoms and an unambiguous glucose reading. Do not let a single number start lifelong treatment without a check, and do not use “it needs confirming” as a reason to disappear for a year.
If the number does not match how you feel, or does not match your home readings, say so directly. Home glucose values that consistently sit far from what the A1c implies are a clue, and the causes above are where a good clinician will look. Bring your readings. A pattern is much harder to dismiss than a recollection.
Frequently asked questions
Do I need to fast before an A1c test? No. That is one of its practical advantages.
How often should it be checked? Every three months if treatment is changing or you are above target, every six if you are stable. More often rarely helps, since the biology has not turned over.
Can A1c be too low? Yes, and it deserves attention rather than congratulation. On insulin or a sulfonylurea, an unexpectedly low result can reflect frequent hypoglycaemia averaging out the highs.
I have sickle cell trait. Is my A1c reliable? Often yes with modern laboratory methods, but this depends on the assay your lab uses. It is a fair question to ask directly, and a good clinician will not mind being asked.
The point
A1c is a good test and an average, and averages hide things. Two people sharing a result of 7.5% can be living completely different lives, one steady, the other swinging between 50 and 300 mg/dL, and only one of them is safe.
So read your number as the strong summary it is, then ask the second question: does anything in your own biology make that summary less trustworthy than it looks? For a lot of people, particularly those carrying haemoglobin variants nobody thought to mention, that is the question that changes the answer.
This article is general health information and is not a substitute for individual medical advice. Discuss your own results with your clinician.