Most of the writing about these drugs — including a good deal on this site — is about weight. For a very large number of people taking them, the number that actually determines their care is HbA1c.
It is also a number people misread constantly, usually by expecting it to respond faster than it can.
What the number is
HbA1c measures the proportion of haemoglobin that has become glycated — sugar bound to it non-enzymatically, in proportion to how much glucose the red cell has been sitting in.
Red cells live about 120 days. So the test integrates glucose exposure over roughly three months, weighted toward the recent past because newer cells outnumber the ones about to be cleared. Roughly half the value is set by the previous month.
Two useful consequences. You cannot fix an A1c in the week before a blood test — a fact that surprises a remarkable number of people who try. And a result taken six weeks after a dose increase is showing you a period that was mostly spent on the old dose.
1.5–1.8
Percentage-point A1c reduction with semaglutide 1.0 mg
SUSTAIN programme
2.0–2.4
Percentage-point reduction with tirzepatide at higher doses
SURPASS programme
120 days
Red blood cell lifespan — why the number lags what you are doing now
The reference points
| A1c | Interpretation |
|---|---|
| Below 5.7% | Normal |
| 5.7 – 6.4% | Prediabetes |
| 6.5% or above | Diabetes range |
| Below 7% | Typical treatment target for most adults with type 2 diabetes |
Targets are individualised rather than universal. Tighter control is generally pursued in younger people with long life expectancy and no hypoglycaemia risk. Looser targets are appropriate in older adults, people with significant comorbidity, and anyone where the cost of a hypoglycaemic episode is high — a consideration discussed further in GLP-1s after 65.
The prediabetes band matters for a second reason: it is a qualifying comorbidity for weight-management prescribing at a BMI of 27 or above, and a great many people in that band have never been tested. See do you qualify.
Why your drop may differ
Starting A1c is the biggest factor. Trials consistently show larger absolute reductions in people who start higher. Someone at 10.5 percent has considerably more room to fall than someone at 7.2, and comparing your drop to a headline average without accounting for that is misleading.
Dose. The relationship is real across the ladder — see the GLP-1 dose charts.
Duration of diabetes. Longer-standing diabetes generally means less residual beta cell function and a smaller response.
What else you are taking. Metformin, SGLT2 inhibitors, insulin and sulfonylureas all contribute, and the GLP-1’s incremental effect is measured on top of them.
Weight loss. The two effects are related but not identical, and this is worth knowing: people sometimes get a good A1c response with modest weight loss, or vice versa.
When the number lies
A1c assumes red cells behave normally. Several common situations mean they do not.
Iron deficiency falsely raises it. Older red cells accumulate more glycation, and iron deficiency extends cell lifespan. This matters here specifically: iron deficiency is common in people eating a third of their previous intake, and more common again in anyone menstruating — see what happens to your cycle.
Recent blood loss or haemolysis falsely lowers it, by populating the blood with young cells.
Chronic kidney disease distorts it in both directions.
Haemoglobin variants — sickle cell trait, thalassaemia and others — interfere with some assay methods.
Pregnancy changes red cell turnover substantially.
If your A1c does not match your glucose readings, one of these is usually why, and the answer is to look rather than to assume one of the two numbers is wrong. Fructosamine or continuous glucose monitoring can settle it.
You cannot revise for this test. The number was decided over the last three months, and mostly over the last one.
What A1c does not tell you
It is an average, and averages conceal.
Two people with identical A1c values can have completely different glucose profiles — one steady, one swinging between hypoglycaemia and high post-meal peaks. The second person is doing considerably worse, and the single number will not show it. Continuous glucose monitoring, where available, adds time-in-range and variability, which are increasingly what clinicians want to see.
It also says nothing about the outcomes people actually care about. The cardiovascular benefit and the kidney result in this drug class were not fully explained by glycaemic control, and appeared earlier than glucose changes alone would predict.
Practical points
Test at three-month intervals while adjusting, then six-monthly once stable. What else belongs on the panel is in bloodwork on a GLP-1.
Do not test six weeks after a dose change and conclude anything. You are reading the previous dose.
Ask for the actual number, and keep the sequence. A trend across four results tells you far more than any single value, and you are the only person guaranteed to still have all four — the same argument as in tracking your GLP-1 journey.
Do not chase a target through hypoglycaemia. An A1c of 6.4 achieved with regular lows is worse than 7.1 achieved steadily, particularly in older adults.
The summary
Expect one and a half to two points from semaglutide, a little more from tirzepatide, more than that if you started high, and less if your diabetes is long-standing.
Expect to see it at three months rather than three weeks — and expect your other diabetes medication to need reducing as it falls, which is the part that requires somebody to be paying attention.