1. The band, and what it is called
Between a normal blood sugar and diabetes there is a range with several names and one meaning.
| Test | Normal | The band | Diabetes |
|---|---|---|---|
| HbA1c | Below 42 mmol/mol | 42–47 mmol/mol | 48 or above |
| Fasting glucose | Below 5.5 mmol/L | 5.5–6.9 mmol/L | 7.0 or above |
The NHS calls it non-diabetic hyperglycaemia. Most people call it prediabetes. The two mean the same thing, and the official term is deliberately duller because the popular one implies an inevitability the evidence does not support.
Being in the band is not a diagnosis of a disease and it is not a sentence, and there are numbers for that rather than reassurance.
In the matched-cohort evaluation of the English programme, about 15% of people in the band who were not referred had developed type 2 diabetes within three years, against about 13% of those who were. So roughly six in seven did not convert over three years either way — and the difference referral made was a few percentage points, which is the honest size of the effect.
And it runs both ways at older ages. In people over 70, progression is uncommon and reversion to normal glucose is at least as likely as progression. Where you sit in the band matters too: 46 is a different proposition from 42.
What it is is the most actionable result on a routine blood test, and the one most often filed as “slightly raised, we'll keep an eye on it” with nothing attached. This chapter is about what should be attached.
2. Why this sits in a cardiovascular guide
Because the damage is not waiting for the diagnosis. Cardiovascular risk rises across this band rather than switching on at 48, and the same insulin resistance behind it drives the pattern described in chapter 2 and the liver changes covered elsewhere: raised triglycerides, low HDL, small dense LDL particles, and a “normal” LDL-C that understates the particle count.
So an HbA1c of 45 is not only a diabetes number. It is a reason to look at the lipid panel differently, and one of the clearest cases on this site of a single result changing how another one should be read.
3. The programme most people in the band have never been offered
England runs Healthier You: the NHS Diabetes Prevention Programme, free, and it reached universal coverage across the country in just over two years. By March 2022 more than a million people had been referred.
- Who is eligible: adults 18 or over, not pregnant, with HbA1c 42–47 mmol/mol or fasting plasma glucose 5.5–6.9 mmol/L recorded within the last 12 months, and no diabetes diagnosis.
- And a route people miss entirely: a history of gestational diabetes qualifies you even with a normal HbA1c. If you had diabetes in pregnancy, you are eligible on that basis alone.
- What it is: nine months of structured behaviour-change support — at least 13 group sessions and a minimum of 16 hours of contact, aimed at weight, activity and diet. A digital version is offered as an alternative to the group format.
- An age note: some providers cap routine eligibility around 80 and ask for GP confirmation above it. There is no national ceiling — check your local provider rather than assuming either way.
- How to get on it: referral from your GP practice, and in many areas you can self-refer. You have to ask. A raised HbA1c does not generate an invitation the way a screening programme does.
4. What it achieves, and the number nobody quotes
Among people who complete it, the national service evaluation found average reductions of about 3.6 kg in weight and 2.1 mmol/mol in HbA1c. Modest in isolation, and meaningful when the whole point is to move someone off a trajectory rather than to treat a disease.
Then the funnel, which is the most useful thing on this page. Of 324,699 people referred in the first phase:
- 53% attended the initial assessment
- 36% attended at least one group session
- 19% completed the programme — where “completed” means attending more than 60% of sessions, which is the definition used in the published service evaluation
Four in five people referred did not finish. That is not a criticism of the participants: nine months and sixteen hours is a real ask of someone who feels perfectly well and has been told their blood test is slightly off.
And the population-level result, which is what the funnel produces. Referral was associated with a reduction in conversion of a few percentage points over three years — the 15% against 13% in section 1. That is what a programme with a 19% completion rate delivers across everybody referred, and stating the two together is what makes the next paragraph an evidence-based argument rather than encouragement.
The useful conclusion is the opposite of discouraging. The intervention works for people who do it, and the binding constraint is finishing rather than efficacy. So the question worth asking at referral is not “does this work” but “which format am I actually going to complete — the group one or the digital one?” Choosing on that basis is more likely to change your result than anything else about the programme.
5. What to do if there is no programme, or you will not finish one
The trial evidence behind diabetes prevention programmes worldwide is strong and consistent, and what it rests on is not mysterious.
- Weight, and specifically visceral fat. The single largest lever. See the body composition tool, which uses waist-to-height rather than BMI for the reason set out there.
- Resistance training, not only walking. Muscle is where most glucose is disposed of, and more of it means more places for glucose to go. Covered in the exercise chapter.
- Activity within about an hour of eating. A short walk after a meal blunts the glucose rise more than the same walk taken earlier — free, under-used, and worth knowing that what has been measured is the glucose curve, not the diabetes. Graded Moderate for postprandial glucose and ungraded for progression, which nobody has tested.
- Liquid calories first. The highest-yield single dietary change, and the one that costs least. See the eating topic.
- Sleep and alcohol, both of which move glucose handling more than people expect — sleep and alcohol.
- An annual HbA1c. NICE asks for yearly review of anyone in this band in primary care. If you have been in it for three years and not had a repeat, that is worth chasing.
What this chapter will not do is give you a carbohydrate target. The evidence for any specific macronutrient split over another for prevention is far weaker than the evidence for the things above, and a number here would imply a precision that does not exist.
6. If you have already crossed the line
Different chapter, different evidence, and worth knowing exists: type 2 diabetes can go into remission in a meaningful proportion of people through substantial weight loss, which was demonstrated in a randomised UK trial using a total diet replacement approach. The NHS runs a programme on that basis for people diagnosed within recent years.
The numbers, with the gradient that explains them. In the DiRECT trial remission was about 46% at one year and 36% at two, falling to roughly 13% at five years — and among those who lost 15 kg or more, 86% at one year. The gradient is the finding: this tracks weight lost and weight kept off, not the diet used to lose it.
To be clear about scope: DiRECT and the NHS programme are for people already diagnosed with type 2 diabetes. On this page they are a bridge, not an offer — you cannot be referred on a prediabetes result.
Remission is not cure — the tendency remains and weight regain returns the diabetes — and it works best closest to diagnosis. It is mentioned here because a great many people diagnosed with type 2 diabetes are never told it is possible, and because it makes the argument for acting in the 42–47 band rather than waiting: the same effort is more effective earlier.
7. The short version
- 42–47 is a result to act on, not to monitor.
- Ask about the NHS programme — it is free, it is national, and you have to ask for it.
- If you had gestational diabetes, you are eligible regardless of your HbA1c.
- Pick the format you will finish, because completion is the binding constraint.
- Read your lipid panel differently once you know you are in this band.
- Get it repeated annually.