Diabetes & Blood Sugar · Prediabetes

Prediabetes: Range, Symptoms and Your Odds of Reversing It

Prediabetes means your blood sugar sits past normal but short of the diabetes line. That is 100 to 125 mg/dL fasting, 5.7 to 6.4% on HbA1c, or 140 to 199 mg/dL at the two-hour mark of a glucose drink.¹ In one Chennai cohort, 58.9% of adults with it had diabetes nine years on.² The odds move with what you do next.

What is prediabetes? Copy link

Prediabetes is a blood sugar result that is too high to call normal and not high enough to call diabetes. The American Diabetes Association defines it by any one of three tests. Fasting glucose, HbA1c, or the two-hour value after a standard 75 g glucose drink.¹ Some reports still print the older term, borderline diabetes. Same thing. Same bands.

It is common here. Very common. The ICMR-INDIAB survey, which tested 113,043 people covering 31 states and territories between 2008 and 2020, put the national prevalence of prediabetes at 15.3% and of diabetes at 11.4%.³ In many states with a lower human development index, people with prediabetes outnumbered people with diabetes. The authors read that as an epidemic still rising in those states.³

The label describes a band of risk. The ADA's own footnote to the criteria says risk is continuous. It extends below the lower limit of the range and climbs steeply toward the upper end.¹ An HbA1c of 6.3% carries a different outlook from 5.7%, even though both get the same word on the report.

What is the prediabetes range? Copy link

Three tests define it. One is enough. Fasting glucose 100 to 125 mg/dL. HbA1c 5.7 to 6.4%. Two-hour glucose after the 75 g drink 140 to 199 mg/dL.¹ The World Health Organization starts the fasting band at 110 instead of 100. The ICMR guideline of 2018 prints both sets side by side.^1,9 Your lab report will usually follow the ADA numbers, the same lines that define a normal sugar level outside pregnancy. Check which it printed.

Is a single result enough? For prediabetes, one abnormal value places you in the band. For diabetes, no, because unless sugar is unequivocally high the ADA requires two abnormal readings, either two different tests on the same day or the same test repeated on another day.¹ A result sitting near a line should be repeated in three to six months.¹

What about sugar after eating? The only after-eating prediabetes range is the two-hour value on the glucose drink, 140 to 199 mg/dL.¹ A reading on a home glucometer two hours after lunch has no prediabetes cut-off attached to it. Meals are not standardized. The tests also disagree with each other. The two-hour test finds more people with prediabetes than fasting glucose or HbA1c do.¹

A fourth number is newer. In 2024 the International Diabetes Federation proposed the one-hour value on the same drink. A result of 155 mg/dL or more marks intermediate hyperglycemia, and 209 mg/dL or more is diabetic range. The federation's case is that the one-hour value predicts future diabetes better than fasting glucose, HbA1c or the two-hour value, and picks people up earlier.¹⁰ The ADA has not adopted it, so most reports will not show it.¹ If your OGTT draws a sample at 60 minutes, ask for the number.

What are the symptoms of prediabetes? Copy link

Usually none, which is as you would expect when a lab number defines the condition. Prediabetes shows up on a blood test long before it shows up in how you feel.^1,4 The classic signs of high sugar are passing urine often, unusual thirst and unexplained weight loss. The ADA uses them to recognize diabetes-range glucose. A random reading at or above 200 mg/dL with those symptoms is diabetes on its own.¹ The symptoms by themselves prove nothing, since thirst and frequent urination have other causes, so arrange a glucose test this week instead of assuming the result will land in the prediabetes band. Severe illness, dehydration or vomiting alongside them needs same-day medical attention.

One visible sign is worth checking. Look at your neck. Darkened, velvety skin on the back and sides of the neck or in the armpits, sometimes with many small skin growths, is called acanthosis nigricans. MedlinePlus lists it as the one sign some people with prediabetes do have. The ADA counts it among the clinical markers of insulin resistance that justify testing.^4,1 It reflects the underlying insulin resistance more than the sugar level itself, so it can be present with a normal report and absent with a high one.

People search for separate symptom lists for women and men, but no guideline has one, because what differs by sex is risk, and polycystic ovary syndrome and a past pregnancy with gestational diabetes both put a woman on the ADA's testing list.¹

Who should get tested, and how often? Copy link

Earlier than you might expect, if you are Indian. The ADA says to test adults of any age who have overweight or obesity plus one more risk factor. For people of Asian ancestry the weight line is a BMI of 23 instead of the usual 25.¹ A diabetic parent or sibling, high blood pressure, low HDL cholesterol, physical inactivity, a past heart attack or stroke, PCOS, fatty liver or the darkened neck skin described above each count as that extra factor.¹ Everyone else starts at 35.¹ The ICMR guideline goes further for India and recommends screening every adult over 30, and earlier if you carry a risk factor.⁹

How often depends on the last result: yearly if you already have prediabetes, at least every three years if the result was normal, and at least every one to three years for life after a pregnancy with gestational diabetes.^1,11

Which test? Fasting glucose and HbA1c are easier, so they are what most people get, but the glucose tolerance test is more sensitive and sorts you into a type of prediabetes, which matters for what comes next.¹ That point gets its own section below.

Can prediabetes be reversed? Copy link

Yes. It also reverses on its own more often than people assume. A 2012 review in The Lancet put it plainly. Each year, 5 to 10% of people with prediabetes progress to diabetes, and about the same proportion go back to normal glucose.¹² Those are broad estimates, and the rates shift with the population studied, the definition used and the starting risk. Reversal here means one thing. A repeat test that falls back under the lines. The ADA standards call this regression to normal glucose, and the risk afterwards is lower, never zero.¹¹

Getting back to normal, even once, changes the long-term odds. In the DPP Outcomes Study, the follow-up of the American prevention trial, people who had returned to normal glucose at least once during the trial went on to have a 56% lower risk of diabetes than people who stayed in the band. The treatment group made no difference to that. It is an association within the follow-up; it does not prove that getting back to normal caused the whole of the difference.¹³ The ADA cites the same finding as a reason to aim for a normal result and to keep aiming for it.¹¹

Now the untreated course. In India it is steep. In the Chennai Urban Rural Epidemiology Study, 1,376 adults who had normal sugar or prediabetes at the start were traced and re-tested after a median of 9.1 years, using the same ADA criteria as this page. Of those who started with prediabetes, 58.9% had diabetes, 176 of the 299 re-tested, and 41.1% did not. Keep that number, and keep its scale in mind: it is one cohort's average, and your own odds depend on which tests are raised and by how much. Of those who started normal, 19.4% had diabetes, another 25.7% had prediabetes, and 54.9% were still normal (arithmetic from the paper's figures).² The authors conclude that Asian Indians have one of the highest incidence rates of diabetes recorded, with rapid conversion from normal to abnormal sugar.²

One more reason not to wait for the diabetes label. In a US cohort followed for about 30 years, prediabetes carried a modestly higher risk of heart disease, heart failure, kidney disease and death. Of that excess risk, 85% remained after the researchers accounted for who went on to develop diabetes.¹⁴ The authors read that as prediabetes contributing to those outcomes over and above the diabetes that may follow it. It is an observational finding in an older American population. Shared factors such as weight, insulin resistance, blood pressure and lipids could explain part of the link, so it shows an association and does not prove that mildly raised glucose causes the harm on its own. The direction matches the Da Qing follow-up the ADA cites. People who regressed to normal, or stayed with impaired glucose tolerance, had less cardiovascular and microvascular disease over 30 years.¹¹

How much do weight loss and diet actually help? Copy link

A lot in the best trials. Less in the real-world versions. The US Diabetes Prevention Program randomized 3,234 adults with prediabetes to an intensive lifestyle program, metformin or placebo, and over three years the lifestyle program cut new diabetes by 58% and metformin by 31%.¹⁵ The program's two goals were losing at least 7% of body weight and doing 150 weekly minutes of moderate activity such as brisk walking.¹¹ Weight loss did most of the work: each kilogram lost was associated with a 16% lower risk of progression over 3.2 years.¹¹

Does the benefit last? Mostly, yes. The 21-year follow-up, published in 2025, found that the lifestyle group still had a lower incidence of diabetes, with a hazard ratio of 0.76. They gained a median of 3.5 extra diabetes-free years. Almost all of that came from the first three years. The effect was banked early and did not grow with time.¹⁵ An August 2026 analysis of the 1,173 participants with Medicare records found that the lifestyle group also went on to accumulate fewer chronic conditions in later life. The metformin group did not differ from placebo.¹⁶

Indian trials have been run. The numbers are more modest. In the Indian Diabetes Prevention Program, 531 adults whose two-hour glucose was raised, mean BMI 25.8, were randomized to a control group, lifestyle advice, metformin or both. Over a median of 30 months, the three-year cumulative incidence of diabetes was 55.0% among controls, 39.3% with lifestyle advice, 40.5% with metformin and 39.5% with both.⁵ That is a relative risk reduction of 28.5% for lifestyle, about half the US figure. The investigators describe their participants as younger, leaner and more insulin resistant than the American, Finnish and Chinese trial populations.⁵ Treating six or seven people prevented one case of diabetes.⁵ Every arm beat the control group.

Two later Indian trials bracket it. D-CLIP in Chennai gave 578 overweight adults a six-month, culturally adapted version of the US curriculum, with metformin added for those at highest risk. Three-year diabetes incidence fell from 34.9% to 25.7%, which the trial reports as a 32% relative risk reduction from its hazard model.¹⁷ Note that 72% of the program group ended up needing metformin as well.¹⁷ The Kerala Diabetes Prevention Program tried a cheaper route, 15 group sessions mostly led by trained lay peers at a cost of US$22.5 per person. At two years 14.9% of the program group had diabetes against 17.1% of controls, a gap that did not reach statistical significance.¹⁸ Light-touch advice improved fruit and vegetable intake and physical functioning scores. It did not reliably stop diabetes. Cheap is not the same as effective.

What does a prediabetes diet plan look like?

Less restrictive than you fear, and less specific than the internet promises. The ADA's position is that no ideal split of carbohydrate, protein and fat has been shown to prevent diabetes. Mediterranean, plant-based, DASH and low-carbohydrate patterns can all work, and what they share is a base of whole grains, legumes, nuts, fruit and vegetables with little refined or processed food.¹¹ The one Indian dataset that models this is ICMR-INDIAB, whose optimization model, built on the diets of 18,090 adults, recommends 54 to 57% of energy from carbohydrate and 16 to 20% from protein to prevent progression from prediabetes. For most of the survey population that means less carbohydrate and more protein than they currently eat.¹⁹ That is a statistical model fitted to survey data. It has not been tested in a trial, so take the direction from it and leave the exact plate to your dietitian. Legumes, dairy, eggs, fish and soy all raise the protein share. The foods to avoid are the ones every pattern above drops, which are refined grains and processed food. Whole fruit stays on the list.¹¹

What about a 7-day meal plan for prediabetes? No particular seven-day menu has been shown to be the standard for preventing diabetes, and the programs that worked did not hand out one. The US program used individual calorie and fat goals, self-monitoring and a 16-session curriculum to get people to at least 7% weight loss, with 150 weekly minutes of brisk-walking-level activity alongside.¹¹ A plan built from what you already eat, with the carbohydrate share trimmed and protein raised, is closer to the evidence than any printed week. As an example of the direction, and nothing more: two rotis instead of three, a full bowl of dal, a bowl of vegetables and a bowl of curd. That lunch moves the split the way the model points. No trial has tested that plate.

Does exercise for prediabetes work without weight loss?

It helps even when the scale does not move. In the US program, participants who reached 150 minutes a week of activity without hitting the weight goal still cut their diabetes risk by 44%.¹¹ Resistance training can count toward the total, and breaking up long sitting lowers the glucose rise after meals.¹¹ Sleep matters too. The ADA now notes a U-shaped relationship. Under 6 hours or over 9 hours a night raises type 2 diabetes risk, including progression from prediabetes, by up to 50%.¹¹

How lifestyle program results differ by prediabetes type Copy link

On the trials' evidence, prediabetes is several conditions. It comes in types, defined by which test is abnormal, and the type changes how much benefit the trials could show from diet and exercise. A 2023 pooled analysis took individual data from four randomized lifestyle trials, 2,794 adults in all. The programs cut new diabetes by 35% in people whose two-hour value was raised, the group doctors call impaired glucose tolerance (hazard ratio 0.65), and by 49% in people with both values raised (0.51). In people whose only abnormality was a raised fasting glucose, the hazard ratio was 0.97, with a confidence interval that spans no effect.⁶

The ADA has drawn the same line. Its 2026 standards state that the prevention trials showed benefit mainly in people with a raised two-hour value. They did not show it in isolated impaired fasting glucose, nor in prediabetes defined by HbA1c alone.¹ D-CLIP in Chennai reported numerically different results by type, 36% for both values raised, 31% for a raised two-hour glucose and 12% for a raised fasting glucose alone, but the differences were not statistically significant (P = 0.77). Because 72% of its participants also took metformin, that trial cannot isolate what lifestyle change alone did in each group.¹⁷ The firmer finding is the pooled one above, and even there a non-significant result in isolated fasting glucose is an absence of proof, which is different from proof that lifestyle change does nothing.

Here is the catch. Fasting glucose and HbA1c are the usual screening tests because they are easy to run, so they are what most people were diagnosed on.¹ Neither one tells you whether your two-hour value is raised. So the result that put you in the band may say little about how much the standard advice can do for you. Odd, but true. A glucose tolerance test with the 75 g drink can clarify that, and it is worth asking about when your fasting glucose and HbA1c disagree, or when knowing the two-hour value would change what your doctor does. The ADA accepts any of the three tests for screening and does not ask everyone with prediabetes to have one.¹ Your doctor can judge whether it adds anything to the results you already hold. If it is done, eat normally for three days before it and fast overnight; the standard test draws blood fasting and at two hours, and a one-hour sample is an optional extra.¹ If the two-hour value is 140 to 199 mg/dL, you are in the group the trials helped most. If only your fasting glucose is raised, the trials have not yet shown how much lifestyle change prevents, so healthy eating, activity and weight management still stand, with closer monitoring and a conversation about whether metformin fits.

How many criteria you meet matters as well. A Korean cohort of 7,186 adults, tested every two years for up to 20 years and published in October 2026, gives the numbers. Meeting one prediabetes criterion raised the risk of diabetes about threefold. Two criteria, about sevenfold. All three, about fifteenfold. People who met all three spent the shortest time in prediabetes, 1.7 years on average, and had a 5.4% chance of going back to normal.²⁰ These are East Asian data, so apply them to Indians with care, but they are the clearest picture yet of why two abnormal values deserve more urgency than one.

Can iron deficiency make your HbA1c read as prediabetes? Copy link

Yes. In India this is not a footnote. Anemia is why it matters. HbA1c measures glucose stuck to hemoglobin over the roughly 120-day life of a red blood cell, so anything that changes how long red cells live, or changes the hemoglobin itself, changes the number without changing your sugar.¹ The direction depends on the cause, and anemia is a broad word: it has many causes, and iron deficiency is only one of them. A 2015 systematic review of 12 studies in people without diabetes found that iron deficiency, with or without anemia, raised HbA1c compared with controls while glucose stayed the same. Anemias from other causes may push it the other way, downward.⁷ NFHS-5, the government's household health survey, put anemia at 57% among women in the 15 to 49 age band.⁸

The guideline answer is simple. Switch tests. The ADA recommends plasma glucose criteria for diagnosis in conditions that alter red-cell turnover, which includes anemia and iron status. It says HbA1c may not be suitable in people with anemia at all.¹ Hemoglobin variants, which some assays misread, are a separate interference listed in the same table.¹ The ICMR guideline notes that HbA1c has limitations as a screening test in India without spelling them out.⁹

So if your HbA1c sits between 5.7 and 6.4% and your hemoglobin is low, ask for a fasting glucose or a glucose tolerance test before you accept the label. A normal hemoglobin does not rule iron deficiency out, so where it is plausible your doctor may check ferritin as well. When HbA1c may be unreliable, the diagnosis rests on laboratory plasma glucose and the HbA1c is set aside.¹ The ADA also flags the reverse: when glucose and HbA1c disagree consistently, the laboratory should look for an interference before anyone acts on the higher number.¹ A single HbA1c in an iron-deficient woman can start years of worry over a number that was never about sugar.

Should you take metformin for prediabetes? Copy link

For some people, yes. The guidelines are specific about whom. The ADA says to consider metformin in adults at particularly high risk of type 2 diabetes. The groups more likely to benefit in the trials were younger adults aged 25 to 59, people with a BMI of 35 or more, and those with a higher fasting glucose (110 mg/dL or more) or a higher HbA1c (6.0% or more). Women who have had gestational diabetes are on the list too.¹¹ These are risk features your doctor weighs together; nobody has to tick every one. Before prescribing, and during treatment, your doctor will check your kidney function, because metformin is usually avoided in kidney disease, and the common side effects are stomach discomfort, diarrhea and nausea.²¹ The ICMR guideline frames it for India. When lifestyle measures are not enough, especially with both fasting and two-hour values raised, metformin may be considered in Indians who progress rapidly.⁹

What does it do? Less than lifestyle change, which comes first, of course. In the US program metformin cut new diabetes by 31%, about half the lifestyle effect, and over 21 years it added a median of 2.5 diabetes-free years.¹⁵ In the Indian program it performed the same as lifestyle advice, 40.5% against 39.3%, and giving both together to the fourth group added nothing, at 39.5%.⁵ The ADA notes that no medicine is approved by the US regulator for the sole purpose of preventing type 2 diabetes. Any drug used this way probably needs to be taken long term, because the effect fades when it stops.¹¹ People on metformin for years should have vitamin B12 checked periodically, more so if they follow a vegan diet or have anemia or tingling in the feet.¹¹

What about the weight-loss injections? The ADA's position is that GLP-1-based medicines for weight management can be considered when overweight or obesity is driving the risk. It also notes there are no long-term data to support any drug other than metformin for prevention alone.¹¹ In someone who meets the criteria for obesity treatment, a GLP-1-based medicine can improve weight, glucose regulation and the insulin resistance behind both, and lower the chance of progression. Prediabetes on its own does not justify a prescription, and the long-term prevention evidence is stronger for metformin.¹¹

How long does reversing prediabetes take? Copy link

No trial gives a date. Be wary of anyone selling a 30-day fix. Weeks for glucose, months for HbA1c. Plan around that. Glucose can change within weeks of eating and moving differently. HbA1c cannot, because it averages the preceding two or three months, which is why the ADA suggests repeating a borderline result after three to six months.¹ The US program's core curriculum ran for six months at a weight-loss pace of 0.5 to 1 kg a week. The largest share of its 21-year benefit was created in the first three years.^11,15

So plan it this way. Change what you eat and move more now, retest at three months to see which way the number is moving, and then test yearly for as long as you remain in the band, which is what the ADA asks for.¹¹ An HbA1c that drifts from 6.1 to 5.8 is progress even though both are in the prediabetes range, because risk follows the number itself.¹

Prediabetes and pregnancy Copy link

If you have had gestational diabetes, you already sit on the ADA's list for testing every one to three years.¹ If you have prediabetes and are planning a pregnancy, tell your obstetrician early, because pregnancy is screened and managed on its own glucose thresholds.¹ Abnormal glucose found during a pregnancy is read on pregnancy-specific criteria by the obstetric team. Before 15 weeks, the ADA treats an HbA1c between 5.9 and 6.4%, and likewise a fasting glucose between 110 and 125 mg/dL, as a marker of higher risk. The risk is of pregnancy complications and of a later gestational diabetes diagnosis. That differs from a diagnosis of gestational diabetes itself, and it is a reason to act sooner than the routine test at 24 to 28 weeks.¹ The ADA's pregnancy section suggests nutrition counselling and periodic fasting glucose checks in the meantime, with any treatment a shared decision. The benefit of treating early abnormal glucose is still uncertain, and professional guidance differs sharply.²²