HbA1c: What Your Blood Sugar Average Actually Tells You

An Average, Not a Snapshot

Your HbA1c is a three-month weighted average of blood glucose. That single fact explains most of what people get wrong about it.

It is not a reading. It cannot be gamed by fasting. And it hides everything interesting about the shape of your glucose curve.

Here is what it measures, where the UK lines sit, the four situations where it lies to you, and what the supplement trials found. On that last point, our own category does not come out well.

The Chemistry, Briefly

Glucose sticks to haemoglobin spontaneously. No enzyme involved. The more glucose in circulation, the more of your haemoglobin ends up glycated, and once it happens it is permanent for the life of that red cell.

Red cells last around 120 days. So your bloodstream holds cells of every age, each carrying a chemical record of the glucose it swam in. Measure the glycated fraction and you get an average.

The weighting is the part people miss. Recent exposure counts for more, because older cells are steadily being cleared. The most recent month drives roughly half the result. The two months before that make up the rest.

Practical consequence: a clean fortnight before your appointment moves the needle slightly. It does not rewrite the number.

The UK Thresholds

The UK reports in mmol/mol under the IFCC standard. American sources and most supplement marketing still use the older DCCT percentage scale, which is why the same result appears as two different-looking numbers.

• Below 42 mmol/mol (below 6.0%) — normal.
• 42 to 47 mmol/mol (6.0 to 6.4%) — non-diabetic hyperglycaemia. Prediabetes, colloquially.
• 48 mmol/mol or above (6.5%+) — meets the NICE threshold for type 2 diabetes, confirmed on a repeat unless symptoms are present.

These do not move with age or sex. A 70-year-old and a 35-year-old at 44 mmol/mol are classified identically.

Worth knowing: 48 is not a biological cliff edge. It was chosen because retinopathy risk starts climbing steeply around there in population data. Below it, risk does not disappear — it declines gradually. Someone at 46 is not "fine" in the way someone at 32 is fine.

The categories are administratively useful and biologically fuzzy. Treat them accordingly.

Translating It: Estimated Average Glucose

Because mmol/mol is abstract, clinicians sometimes convert to estimated average glucose. The conversion comes from Nathan and colleagues in 2008, who ran continuous glucose monitoring against HbA1c in several hundred people.

Roughly: 42 mmol/mol maps to an average glucose near 7.0 mmol/L. 48 maps to about 7.8 mmol/L.

The correlation is strong across a population and looser for any individual. That is why eAG never replaced the raw number.

Four Ways HbA1c Lies

The test assumes red cells live a normal lifespan and haemoglobin behaves normally. Break either assumption and the number stops meaning what it appears to mean.

Short-lived red cells read low. Haemolytic anaemia, recent blood loss, a recent transfusion, chronic liver disease. Fewer old cells in circulation means less accumulated glycation, and a falsely reassuring result.

Long-lived red cells read high. Iron deficiency and B12 deficiency both extend average cell age. Either can push HbA1c up several mmol/mol with no change in glucose whatsoever. This matters more than it sounds — untreated iron deficiency can get someone labelled prediabetic on the strength of their ferritin.

Haemoglobin variants interfere with the assay. Sickle cell trait, and haemoglobin C, D and E variants, are common in people of African, Mediterranean, Middle Eastern and South Asian ancestry. Depending on the lab method, results shift either way. Most UK labs handle common variants, but it is a fair thing to raise if it applies to you.

Sometimes it simply does not apply. WHO and NICE exclude HbA1c in pregnancy, in suspected type 1 diabetes, in under-18s, and in anyone with rapid-onset symptoms. Fasting glucose or an oral glucose tolerance test is used instead.

If your result surprises you, check iron status before you change anything else.

What the Average Conceals

Two people can share an HbA1c of 40 mmol/mol and live completely different metabolic lives.

One sits at 5.5 mmol/L all day. The other swings from 4.0 to 9.0 after every meal. Same average. There is decent observational evidence that glycaemic variability carries risk of its own, independent of the mean — though it is far less established than the mean itself.

This is the honest case for a CGM as a complement rather than a replacement. HbA1c gives you the centre of gravity. A CGM gives you the shape. If your number is drifting upward and you want to know why, the shape is usually where the answer is.

For most people, neither is urgent. For someone already training hard and eating well whose HbA1c is creeping anyway, the shape is the interesting question.

What Actually Moves It

The levers with real effect sizes behind them are the boring ones. You already know most of this list, which is rather the point — nothing here is a product.

• Dietary pattern, particularly refined carbohydrate and ultra-processed food volume.
• Visceral fat, which matters more than total body weight.
• Resistance training, because muscle is where glucose gets disposed of.
• Aerobic work, which improves insulin sensitivity both acutely and chronically.
• Sleep. Even short-term restriction measurably reduces insulin sensitivity in controlled studies.

We ordered these by effect size in our nutrition strategies for healthy ageing after 40, and the ordering was deliberate.

The Supplement Evidence, Straight

This is where we could be commercially convenient. We are not going to be.

NMN does not lower HbA1c. A 2024 systematic review pooled eight randomised controlled trials — 342 middle-aged and older adults, 250 to 2,000 mg a day, two to twelve weeks. No significant effect on HbA1c, fasting glucose, fasting insulin, HOMA-IR or lipids.

That is the headline and it deserves to be. There is a more interesting result underneath, but it does not overturn the conclusion.

Read what that did and did not show. It measured muscle insulin sensitivity under clamp conditions — a laboratory measure — in 25 women in one specific metabolic state. HbA1c did not change. It has not been replicated at scale.

Presenting it as evidence that NMN improves blood sugar generally is a misreading. You will see that misreading constantly. For what NMN does have evidence behind it, we set the case out in what NMN actually is and how it works.

Omega-3 is mixed rather than null. A 2018 meta-analysis and meta-regression of 45 randomised trials in 2,674 people with type 2 diabetes found a small but statistically significant HbA1c reduction — effect size −0.27 — alongside more consistent triglyceride reductions. A separate meta-analysis of 20 trials found no meaningful HbA1c change at all.

The defensible reading: any glycaemic effect is small, inconsistent, and studied mainly in people who already have type 2 diabetes rather than in healthy adults. The triglyceride effect is far better evidenced. Our algae-derived Omega-3 DHA is formulated with that literature in mind, not a glycaemic claim.

Getting Tested in the UK

HbA1c is not universally screened, but it is part of the NHS Health Check offered every five years to adults aged 40 to 74 in England. It also gets added routinely wherever there is a metabolic reason to look.

Ask for it explicitly if you have a family history of type 2 diabetes, carry central weight, have PCOS, or previously had gestational diabetes. Do not wait for it to appear.

Private testing is cheap and widely available, by pharmacy or postal finger-prick. Two cautions. Different labs and methods are not perfectly interchangeable, so comparing a private result against an NHS one introduces a difference that has nothing to do with your metabolism — track like against like. And a private result crossing a diagnostic threshold still needs confirming through your GP.

Whichever route, get the figure rather than a verbal "it's fine". The number and its direction are the useful part.

What To Do With Your Result

Below 42: note it and look at the trend across previous results. Direction beats category.

42 to 47: NICE recommends lifestyle intervention and annual monitoring, with referral to the NHS Diabetes Prevention Programme available in England. This is the range with the most leverage. It is also the range where you should ask whether iron status is inflating the figure.

48 or above: that is a conversation with your GP, not with a supplement website.

In every case, retest no sooner than three months. Anything faster measures noise.

The Short Version

HbA1c is a good test that gets over-read. It measures a weighted three-month glucose average, it distorts whenever red cell lifespan is abnormal, and it says nothing about the variability underneath. The UK's 42 and 48 are useful administrative lines drawn across a continuous risk gradient.

On supplements: pooled randomised evidence does not support NMN for lowering HbA1c, and the omega-3 signal is small and inconsistent. Diet, visceral fat, training and sleep are where the effect sizes live.

We would rather tell you that than sell you something the trials do not support.

References

1. National Institute for Health and Care Excellence (2024). "Type 2 diabetes: prevention in people at high risk." NICE Public Health Guideline PH38.
2. World Health Organization (2011). "Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus." WHO/NMH/CHP/CPM/11.1.
3. Yoshino, M. et al. (2021). "Nicotinamide mononucleotide increases muscle insulin sensitivity in prediabetic women." Science, 372(6547), 1224-1229. PMID: 33888596
4. Zhong, O. et al. (2024). "Effects of Nicotinamide Mononucleotide on Glucose and Lipid Metabolism in Adults: A Systematic Review and Meta-analysis of Randomised Controlled Trials." Current Diabetes Reports. PMID: 39531138
5. O'Mahoney, L. L. et al. (2018). "Omega-3 polyunsaturated fatty acids favourably modulate cardiometabolic biomarkers in type 2 diabetes: a meta-analysis and meta-regression of randomized controlled trials." Cardiovascular Diabetology, 17(1), 98. PMID: 29981570
6. Chen, C. et al. (2015). "Effects of Omega-3 Fatty Acid Supplementation on Glucose Control and Lipid Levels in Type 2 Diabetes: A Meta-Analysis." PLOS One, 10(10), e0139565.
7. Nathan, D. M. et al. (2008). "Translating the A1C assay into estimated average glucose values." Diabetes Care, 31(8), 1473-1478.

Evidence First. Even When It Costs Us.

We publish what the trials found, including the null results. Our Omega-3 DHA is algae-derived, third-party tested, and sold on the literature that actually supports it.

See Omega-3 DHA →