136 Million Indians Have Prediabetes: What That Actually Means

The number arrives without instructions. A newspaper reports that 136 million Indians have prediabetes and 101 million have diabetes, and the reasonable next thought is whether your own fasting glucose of 104 mg/dL puts you inside that 136 million or outside it. The answer depends on which rule book was used, and India does not have one. Under the criteria that produced the national figure, 104 is normal. Under the criteria your clinic may well be using, it is not. What follows is where the 136 million came from, which thresholds define prediabetes here, how often Indian adults with the label go on to develop diabetes, and how often they do not.

Key Takeaways:

• The 136 million figure is a 2021 IDF-modelled projection from ICMR-INDIAB survey data collected between 2008 and 2020. It is not a direct count.
• RSSDI uses a fasting glucose cut-off of 100 mg/dL for IFG. WHO uses 110 mg/dL. The same result of 104 mg/dL is prediabetes on one definition and normal on the other.
• CURES found 56.2% of those with combined IFG and IGT had converted to diabetes. IGT without IFG is not low risk: 58.9% converted within five years.
• Being inside the 136 million means a test came back in a particular range on one day. It does not mean conversion is certain or that the label stays fixed.
• The useful questions are not "do I have prediabetes" but "what is my fasting glucose, what is my two-hour result, what is my trend, and what has changed in the twelve months since the last test."

Both figures come from ICMR-INDIAB-17, reported in Lancet Diabetes & Endocrinology in 2023. It is the largest metabolic survey India has run: a community-based, stratified multistage sample drawn across 31 states and union territories, with fieldwork spanning October 2008 to December 2020. Villages and urban census blocks were selected with probability proportional to size, households by systematic sampling, and one adult aged 20 or over per household by the WHO Kish method, targeting roughly 4,000 people per state. Of 113,043 participants, 107,119 had complete oral glucose tolerance test data.

Glycaemic status was measured, not self-reported. Fasting capillary blood glucose was taken after an 8 to 12 hour overnight fast, and every participant without known diabetes then completed a standard 75 g oral glucose tolerance test with a 2-hour post-load reading. WHO criteria were applied, with thresholds adapted for capillary rather than venous samples, which is why the study's impaired glucose tolerance band reads as 160 to 219 mg/dL rather than the familiar plasma figures. The weighted prevalence in that sample was 11.4% for diabetes and 15.3% for prediabetes. Applying those rates to India's projected adult population for 2021 is what produced the headline: 101 million people with diabetes and 136 million with prediabetes, stated in the paper's own text.

That last step is worth holding onto. The 136 million is a projection, not a headcount. The paper describes the population denominator and projection method only briefly, without a breakdown by age band or sex, so the precision implied by a figure like "136 million" is not something the published tables let anyone verify. The prevalence figures are the measured part. The counts are arithmetic performed on them, and both belong to the lancet paper rather than to the ministry releases and news reports that repeat them.

INDIAB-17 used WHO criteria. WHO sets impaired fasting glucose at 110 to 125 mg/dL and impaired glucose tolerance at a 2-hour value of 140 to 199 mg/dL, and does not endorse HbA1c for diagnosing prediabetes at all. The American Diabetes Association sets impaired fasting glucose lower, at 100 to 125 mg/dL, keeps the same 2-hour band, and adds HbA1c of 5.7% to 6.4% as a third qualifying route.

A 10 mg/dL difference in one threshold sounds trivial. It is not. The INDIAB-17 supplementary appendix reports the same survey re-analysed under ADA fasting criteria, and prediabetes prevalence rises from 15.3% to about 32.8%. That is the same people, the same blood samples, the same year, with more than twice as many of them labelled. Nothing about anyone's metabolism changed. Only the line did.

WHO (criteria used to produce the 136 million figure):

• Impaired fasting glucose: 110 to 125 mg/dL
• Impaired glucose tolerance: 140 to 199 mg/dL (2-hour post-load)
• HbA1c: not endorsed for diagnosing prediabetes

ADA:

• Impaired fasting glucose: 100 to 125 mg/dL
• Impaired glucose tolerance: 140 to 199 mg/dL (2-hour post-load)
• HbA1c: 5.7% to 6.4%

RSSDI-ESI (for Indian clinical practice):

• Impaired fasting glucose: 100 to 125 mg/dL
• Impaired glucose tolerance: 140 to 199 mg/dL (2-hour post-load)
• HbA1c: 5.7% to 6.4%

Epidemiological analyses describe the trade-off in the same terms both ways: the lower ADA cut-off captures a much larger group, including many people who go on to return to normal glucose levels on their own, while the higher WHO cut-off identifies a smaller group carrying more progression risk. Neither is wrong. They are calibrated for different jobs, one for finding everyone who might benefit from attention and one for counting who is genuinely on the path.

This is the part that catches most readers out, because the divergence is not only between Geneva and Virginia. It runs inside Indian clinical practice.

The Research Society for the Study of Diabetes in India and the Endocrine Society of India, in their joint clinical practice recommendations, explicitly endorse the ADA criteria for the Indian context, defining impaired fasting glucose at 100 to 125 mg/dL. RSSDI's own 2022 clinical practice recommendations, however, list impaired fasting glucose at 110 to 125 mg/dL, alongside impaired glucose tolerance at 140 to 199 mg/dL and HbA1c at 5.7% to 6.4%. So Indian professional guidance carries both fasting bands depending on which document is open, while consistently accepting an HbA1c route that the WHO framework behind the national statistic does not use.

The practical consequence is specific. A fasting glucose of 104 mg/dL is normal under the definition that generated the 136 million and prediabetic under the definition an Indian clinic is likely to apply. An HbA1c of 5.9% carries no prediabetes label under WHO criteria and does under RSSDI and ADA criteria. There is no nationally agreed answer on which definition should govern how risk is communicated in general practice, which means the survey label and the clinic label can genuinely disagree about the same person.

The word covers three states, and they do not carry the same risk. Isolated impaired fasting glucose means a fasting value in the intermediate range with a normal 2-hour value. Isolated impaired glucose tolerance is the reverse. Combined means both.

The Indian Diabetes Prevention Programme separated them and followed 845 adults with impaired glucose tolerance for three years. In the control arms, 56.2% of those with combined impaired fasting glucose and impaired glucose tolerance developed diabetes, against 33.6% of those with isolated impaired glucose tolerance. The direction of travel differed too: 32.1% of the isolated group returned to normal glucose tolerance, against 18% of the combined group.

That is close to a two-fold difference in three-year risk sitting behind a single word. A report that says "prediabetes" without saying which pattern was found has left out the part that most changes what happens next, and an oral glucose tolerance test is the only one of the common tests that can tell the two apart, because a fasting sample alone cannot see the 2-hour value.

The Chennai Urban Rural Epidemiology Study followed 1,376 adults for a median 9.1 years, including 299 who had prediabetes at baseline, using glucose tolerance testing throughout. Among those with prediabetes at the start, 58.9% had developed diabetes by the end, a rate of 78.9 per 1,000 person-years. Commentary in the same journal noted this as among the highest conversion rates reported anywhere. A smaller Kolkata cohort of 144 adults followed for around 32 months found a closely similar rate.

The other half of the CURES result gets quoted far less often. In the investigators' own follow-up response, 17.4% of the participants with baseline prediabetes returned to normal glucose tolerance over the period, and 23.7% were still prediabetic at the end. So roughly two in five did not progress. In the ORANGE study, which followed Indian adolescents and young adults for 10 years, the balance tipped the other way entirely: 70.6% of those with prediabetes reverted to normal glucose levels. Age, and how each study ascertained glucose status, plainly matter, and no single Indian conversion rate applies to every adult.

Against Western trials, the comparison points one way but not cleanly. In the placebo arm of the US Diabetes Prevention Program, which enrolled adults with overweight or obesity who had both impaired fasting glucose and impaired glucose tolerance, diabetes incidence ran at about 11 cases per 100 person-years, and intensive lifestyle intervention brought that down to 4.8.

The Finnish Diabetes Prevention Study and other European lifestyle trials report control-arm rates closer to the DPP than to CURES. That supports the common claim that Asian Indians convert faster than most Western populations, but the evidence base is a handful of cohorts with different entry criteria and diagnostic methods, and standardised head-to-head comparison has not been done. The claim holds as a direction and is thin as a precise multiple.

Being inside the 136 million means that on the day of testing, a fasting or 2-hour glucose value sat in WHO's intermediate range without reaching the diabetes thresholds. It is a measurement of one morning, not a diagnosis of a trajectory, and it says nothing on its own about which of the three patterns produced it. It is also rarely the only thing a metabolic workup turns up, and the liver side of that picture, where Indian data show the same pattern of disease at a normal weight.

What the Indian cohort data support is narrower and more useful than the headline. Progression risk is substantially higher when fasting and post-load values are both abnormal than when only one is. Reversion to normal glucose levels happens in a real minority of Indian adults and in a majority of Indian adolescents. And a single fasting number, which is what most people actually have, cannot distinguish between those situations. If you have been handed a prediabetes label, the questions worth putting to a registered medical practitioner are which criteria were applied, whether a 2-hour value was ever measured, and whether the result has been repeated, because a diagnosis and any decision about what follows it are clinical judgements rather than something to settle from a cut-off table.

A prediabetes result that needs reading against your full picture rather than a number range? A Voy clinician can put your fasting glucose, two-hour result and trend together. Book an assessment.

This article is for general information and education only and is not medical advice, diagnosis, or treatment. GLP-1 and other medications referenced are prescription-only and are appropriate only for certain people under the supervision of a qualified clinician. Do not start, stop, or change any medication based on this article. Please consult a registered medical practitioner about your individual circumstances. Information reflects what was available at the time of review and may change.