Prediabetes

Prediabetes is a diagnosis made from a number rather than from a symptom. It describes blood glucose that sits consistently above the normal range without having reached the threshold at which type 2 diabetes is diagnosed.

Any one of three tests can establish it. An HbA1c between 5.7% and 6.4%. A fasting plasma glucose between 100 and 125 mg/dL, called impaired fasting glucose. Or a two-hour value between 140 and 199 mg/dL on a 75g oral glucose tolerance test, called impaired glucose tolerance. The three do not identify the same people. Someone can have a normal fasting glucose and a clearly abnormal two-hour value, and that combination is common in South Asians.

The cut-offs are a convention, and conventions differ. Most Indian practice follows the American Diabetes Association thresholds given above; the World Health Organization begins impaired fasting glucose at 110 mg/dL and treats an HbA1c of 6.0–6.4% as its high-risk range. That explains an experience many people have — a value of 5.8% flagged in bold on one laboratory's report and left unremarked on another's.

Underneath the number is insulin resistance, and two things are happening at once. Muscle, liver and fat tissue respond less readily to insulin, so the pancreas produces more to achieve the same effect. For years that compensation succeeds and glucose stays normal. Prediabetes is the point at which the compensation is measurably falling behind — the demand has risen and the insulin-producing cells have lost some of their reserve. That mechanism, and the tests that describe it, are covered on the separate page for insulin resistance. This page is about what happens once the glucose number itself has moved.

The scale of it in India is difficult to overstate. The ICMR-INDIAB survey published in 2023 estimated roughly 136 million adults with prediabetes alongside about 101 million with diabetes — in several states, more people sitting in this band than beyond it. Most of them do not know, because there is nothing to feel.

Progression is neither inevitable nor rare. International cohorts describe on the order of 5 to 10% of people with prediabetes developing type 2 diabetes each year, with risk rising steeply through the band — 6.2% carries a very different outlook from 5.8%. Indian data suggests faster movement: in the Indian Diabetes Prevention Programme, run in Chennai in adults with impaired glucose tolerance, more than half of the control group had developed diabetes within three years. A proportion of people also return to normal values, and that is a real outcome rather than a consolation.

The band matters before 6.5% because 6.5% is a line drawn for classification, not a line at which biology changes. Cardiovascular risk rises continuously across the prediabetic range; retinal and nerve changes are present in a small but real proportion of people at the moment diabetes is diagnosed, which means they began earlier; and the interventions that work do more here, while insulin-producing capacity is better preserved, than they do later.

The South Asian part of the picture is the same one that runs through fatty liver and PCOS. At a given weight, Indian bodies tend to carry more fat around the organs and inside the liver, less under the skin, and less skeletal muscle. The 2009 Indian consensus set the overweight threshold at a BMI of 23 and obesity at 25, against 25 and 30 in European populations, and put waist action points at 90 cm for men and 80 cm for women. A normal BMI is not reassurance, and the tape measure is often more informative than the scale.

ⓘThis information is educational and not a diagnosis.

Prediabetes is usually silent. It is found on a routine health check, on insurance or pre-employment screening, or while investigating something unrelated. Nothing was missed — there is genuinely nothing to feel at these glucose levels.

Some people, looking back, recognise a pattern. Heavy sleepiness in the hour after a large rice or roti-based meal. Wanting something sweet by mid-afternoon. Hunger returning two hours after eating properly. A waist that has grown over three or four years while the scale has barely moved, which usually means fat redistributing towards the abdomen while muscle is quietly lost. None of these is specific. They are worth noticing as a pattern, not as proof.

What does more work than symptoms is the risk profile, because it tells you when to check the number rather than wait for something to appear. A parent or sibling with type 2 diabetes. Gestational diabetes in a previous pregnancy, or a baby born unusually large. PCOS. Fatty liver on any earlier scan. Raised triglycerides with a low HDL. Blood pressure creeping upward. A waist above 90 cm in men or 80 cm in women. Darkened velvety skin at the back of the neck or in the armpits, which is a response to high circulating insulin rather than a hygiene problem. Long sitting hours with short sleep. Certain long-term medicines, including steroids and some antipsychotics.

Because type 2 diabetes appears a decade or more earlier in Indians than in European populations, Indian guidance sets a considerably younger age for a first check than Western guidelines do, and earlier still where any of the above applies. The practical version: a first-degree relative with diabetes is reason enough to have a number in your file by your late twenties.

The features listed below are of a different kind. They suggest glucose has already risen well beyond this band, or that a complication needs attention, and they warrant being seen promptly rather than at the next convenient appointment.

HbA1c is the usual starting point. It measures the proportion of haemoglobin that has become glycated, reflecting average glucose over the lifespan of a red cell — roughly three months, weighted towards the most recent four to six weeks. It needs no fasting, which is why it fits real life and why it appears in almost every health-check package sold in India.

Its limitations matter here more than elsewhere, because they are common in this country. Iron deficiency raises HbA1c without glucose having moved, and treating the iron deficiency brings it back down; given how widespread iron deficiency is among Indian women, this alone accounts for a number of borderline results. Haemoglobin variants and thalassaemia trait, frequent in parts of eastern, north-eastern and central India, shorten red cell survival and interfere with some assay methods, usually lowering the value. Recent blood loss, a recent transfusion, pregnancy and chronic kidney disease all distort it. Where the number and the clinical picture disagree, a glucose-based test resolves the disagreement.

A single abnormal result in someone without symptoms is confirmed before a label is applied — ideally the same test, at the same laboratory, on a separate day. Method matters too: HPLC-based analysers and immunoassays behave differently in the presence of haemoglobin variants.

Fasting glucose is straightforward and has one practical trap. Cells in a blood sample continue to consume glucose after collection, so a specimen that sits for two hours before reaching the analyser reads lower than the blood did in your arm. Fluoride tubes slow this and prompt processing matters — worth knowing with the home-collection services most of urban India now uses.

The oral glucose tolerance test costs little in reagent and a great deal in time: a fasting sample, a measured 75g glucose drink, two hours in a laboratory. It earns that time in one situation, common here. In many South Asians the earliest abnormality is a high post-meal peak alongside an entirely normal fasting glucose and a borderline HbA1c, which both of the quicker tests smooth away. With a strong family history, previous gestational diabetes, or a picture that does not fit the reassuring numbers, this is the test that settles it.

Because prediabetes rarely travels alone, the first assessment sensibly extends beyond glucose: a fasting lipid profile read for triglycerides and HDL rather than only total cholesterol, blood pressure, waist circumference, and liver enzymes with an ultrasound. Finding fatty liver at the same time is common and changes nothing about the plan — it reinforces it. Fasting insulin is not required to make this diagnosis; once the glucose number has moved, the mechanism is no longer in question.

Repeat testing every six months while changes are being made is enough, with three months reserved for situations where something is actively being adjusted. Between two HbA1c results, a difference of around 0.3% can be produced by assay and biological variation alone. Testing more often than this reliably produces movement, and reliably fails to distinguish it from noise.

The evidence for structured lifestyle change in prediabetes is among the strongest in preventive medicine, and it deserves to be quoted accurately rather than optimistically.

The large trials in the United States and Finland reduced progression to type 2 diabetes by around 58% over roughly three years, and follow-up over the decades since shows the difference persisting long after the programmes ended. The Indian Diabetes Prevention Programme, run in Asian Indians with impaired glucose tolerance, found a smaller relative benefit — in the region of 29% — in a group that was progressing considerably faster to begin with. Both facts belong in the same sentence. The intervention works here; it works against a stronger current, which is an argument for starting sooner and for expecting the work to be ongoing rather than a project with an end date.

Build the destination for glucose. Skeletal muscle takes up the greater part of glucose after a meal, and how much muscle there is — and how recently it was emptied — decides where that glucose can go. Resistance training two or three times a week improves glucose handling independently of any weight change, which is the point for the many people here who have little weight to lose. A session's effect on glucose uptake lasts a day or two, so frequency serves you better than intensity.

Walk after the largest meal. Ten to fifteen minutes of unhurried walking after dinner blunts the post-meal glucose rise. Where impaired glucose tolerance is the abnormality found, this addresses precisely the part that is abnormal, and it fits an Indian evening better than most imported advice.

Weight, where it applies. A sustained reduction of 5 to 7% of body weight had the largest single effect in the prevention trials, and the benefit tracked closely with how much was lost and held. Gradual and maintained outperforms rapid and regained, and a plateau is information about what the plan is asking of you rather than a verdict on you.

Food, inside how Indian households actually eat. The useful change is composition rather than prohibition. A typical plate carries a generous portion of rice or roti, a small katori of dal, one sabzi and some curd — carbohydrate-dominant, with protein arriving in modest quantity. The levers: more protein at each meal from a fuller portion of dal, rajma, chana, sprouts, paneer, curd, soya, eggs or fish; vegetables and salad before the starch rather than after it; rice or roti in a measured portion rather than removed; and coarse grains such as jowar, bajra and ragi through part of the week. Sweetened drinks, packaged juices and the biscuits that accompany four or five cups of tea a day contribute more than their size suggests. Festival weeks are planned for rather than dreaded — a defined portion of mithai, and the walk afterwards, survives longer than a resolution that collapses on day two. In Jain and strict vegetarian households, where onion, garlic and root vegetables are avoided, protein is worth deliberately assembling from dal, moong, paneer, curd and soya rather than assumed. Rice-dominant plates in the south and east and wheat-dominant plates in the north pose the same question: what portion, and what sits alongside it.

Sleep, which is a metabolic intervention. A few nights of short or fragmented sleep measurably reduce insulin sensitivity in healthy volunteers. Where everything else is being done and the number has not moved, sleep duration and untreated obstructive sleep apnoea are the variables most often overlooked. Snoring with daytime sleepiness is worth mentioning to your doctor. Tobacco in every form raises the risk of progression, and alcohol adds carbohydrate, disturbs sleep and loads a liver that in this group is often already carrying fat.

Medication has a defined and limited place. Metformin delayed progression in the prevention trials, with the clearest benefit in younger people, in those with a higher BMI, and in women with previous gestational diabetes. Guidelines list it as an option in specific circumstances rather than as routine practice for everyone in the band. That is an individual decision with your doctor, taken alongside the rest of the plan.

Follow-up is HbA1c at six months, lipids and blood pressure annually, waist every few months, and a plan that is revisited rather than issued once. The most useful review question is not what the number did, but which part of the plan survived contact with an ordinary week — and what to change so that more of it does.

Most of the confusion around prediabetes comes from the report rather than from the condition: what HbA1c measures, why two laboratories comment differently on the same value, and how much of a change is real.

If the more pressing question is why the number moved at all — what insulin is doing, why a waist grows while weight holds steady, and why triglycerides shifted years before glucose did — that mechanism is set out on the page for insulin resistance. Fatty liver, found on so many of the same scans, is the other half of the same picture.

The Learning Session below works through reading your own blood results, HbA1c included, in a small group with time for questions.

If you want this result interpreted alongside your family history, your weight trend and the other reports already in your folder, that belongs in a consultation rather than an article.