"Cut carbs" is the advice handed to almost every Indian adult who comes back from a lab with a borderline sugar reading, and it almost never arrives with a number attached. India now has one.

In a national dietary analysis of 18,090 adults, carbohydrate supplied 62.3% of daily energy and protein 12%, against a guideline benchmark of roughly 15%. Those are statements about the shape of the average plate, not about how much any one person eats. What follows is where the figures come from, what the same dataset found about metabolic risk, and how much weight a national average can honestly carry.

• Across 18,090 adults, carbohydrate supplied 62.3% of daily energy, protein 12% and fat 25.2%.
• Refined cereals such as white rice and refined wheat accounted for 28.5% of energy on their own, with milled whole grains a further 16.2%.
• The "recommended 15%" for protein is the top of ICMR-NIN's 10 to 15% band, not the middle of it.
• Adults in the highest carbohydrate group had higher odds of newly diagnosed type 2 diabetes than those in the lowest, at OR 1.30 (95% CI 1.14 to 1.47). The association is cross-sectional.
• Confidence intervals for the 62.3%, 12% and 25.2% figures themselves sit in the paper's full tables and were not available for this piece.

The figures come from the ICMR-INDIAB dietary profiles analysis, published in Nature Medicine in 2025. Diet was measured in 18,090 adults, a subset of the 113,043 people surveyed by ICMR-INDIAB between 2008 and 2020, using a validated food-frequency questionnaire developed by the Madras Diabetes Research Foundation and scored against EpiNu, an Indian nutrient database.

Five numbers matter, and all five are reported as published rather than derived here:

• Carbohydrate contributed 62.3% of daily energy.
• Protein contributed 12% of daily energy.
• Fat contributed 25.2% of daily energy.
• Inside the carbohydrate share, refined cereals supplied 28.5% of energy.
• Milled whole grains contributed a further 16.2% (the more informative split, because it separates the grain that is polished from the grain that is not).

One absence is worth naming rather than smoothing over. Confidence intervals around those composition means sit in the paper's full tables, which were not accessible for this piece. The intervals around the risk estimates further down are published and are quoted here in full. A mean assembled from a survey that ran across twelve years, on self-reported intake, is a weaker number than three decimal-clean percentages make it look.

The paper measures the 12% protein figure against a national dietary guideline benchmark of around 15% of energy. That benchmark is not a round number picked for contrast. ICMR-NIN's 2020 nutrient recommendations and its 2024 Dietary Guidelines for Indians converge on:

• Carbohydrate: 50 to 60% of energy
• Protein: 10 to 15% of energy
• Fat: 20 to 30% of energy

So the widely quoted "recommended 15%" is the upper edge of a band, not a target sitting in its centre. At 12%, the national average is inside the recommended protein range and near its floor. That is a materially different statement from "Indians eat below the recommended protein intake", which is how the comparison is usually reported.

• Carbohydrate: ICMR-NIN guidance for adults is 50 to 60% of energy; the measured national mean (n=18,090) was 62.3%.
• Protein: ICMR-NIN guidance for adults is 10 to 15% of energy; the measured national mean (n=18,090) was 12%.
• Fat: ICMR-NIN guidance for adults is 20 to 30% of energy; the measured national mean (n=18,090) was 25.2%.

Carbohydrate is the only component that sits outside its band. ICMR-NIN's "My Plate for the Day" model goes further on the source: cereals at no more than 40 to 45% of calories, with pulses and flesh foods at around 17%.

Against that benchmark, What India Eats reports:

• Cereals: about 58% of energy in urban adults and close to 69% in rural adults.
• Separate ICMR-NIN material puts cereals and millets at 51% urban and 65% rural.

The two sets of figures are not identical, and this piece does not resolve which food-group framing each uses.

The single number hides the thing a reader most wants to know, which is whether it describes them. The Nature Medicine tables show carbohydrate shares running higher still in several rice-dominant eastern and southern states and slightly lower in wheat-dominant north-western states.

State-level data from the NSS nutritional intake report for 2011-12 illustrates the spread:

• Rural India overall: cereals supply about 60% of calories
• Urban India overall: cereals supply about 50% of calories
• Punjab, Kerala, Haryana: roughly 46 to 48%
• Odisha, Assam: about 70%

Household consumption survey tables for 2022-24 show cereal calorie shares falling somewhat while often remaining above 45%.

Two things follow. The national composition figure is a midpoint between diets differing by more than twenty percentage points of energy, so it describes a country rather than a household. And the pattern is long-standing rather than newly arrived, which matters when the framing on offer is that Indian eating has recently gone wrong.

The same 18,090 adults were analysed for the association between carbohydrate intake and metabolic outcomes. Compared with the lowest carbohydrate intake group, the highest group had higher odds of four outcomes after adjustment for confounders:

Compared with the lowest carbohydrate intake group, the highest group had:

• Newly diagnosed type 2 diabetes: OR 1.30; 95% confidence interval 1.14 to 1.47.
• Prediabetes: OR 1.20; 95% confidence interval 1.06 to 1.33.
• General obesity: OR 1.22; 95% confidence interval 1.07 to 1.37.
• Abdominal obesity: OR 1.15; 95% confidence interval 1.01 to 1.30.

These are modest effect sizes. Reading them honestly means noticing that the abdominal obesity interval very nearly touches 1. They are also cross-sectional: diet and blood measures were captured at the same point, so they show that a high carbohydrate share and metabolic disease travel together here, not that one produced the other.

How much of that signal belongs to the amount of carbohydrate and how much to the kind of it (polished rice and refined wheat against pulses and coarse grains) is not settled within Indian cohorts.

South Asian studies and reviews describe a consistent metabolic mechanism. High-carbohydrate meals, particularly from refined grains and added sugars, drive:

• Repeated post-meal rises in blood glucose
• Higher circulating insulin
• Raised triglycerides (all three are markers of insulin resistance)

Nothing in that chain is specific to India. What differs is the body it meets.

Adult Asian Indians carry more liver and visceral fat and greater insulin resistance at a lower BMI than White Europeans.

The same energy distribution therefore arrives at a body with less spare capacity to store it safely, and the liver is often where that shows first.

The most direct attempt to answer this used the national data rather than a trial. A 2022 optimisation study in Diabetes Care applied constrained quadratic programming to ICMR-INDIAB to ask which macronutrient mix would minimise HbA1c. The model pointed to lower carbohydrate and higher protein than the current average. It is a modelling exercise on cross-sectional data, and it describes what a guideline might say rather than what any individual should eat.

Intervention evidence from India is thin and short:

• In 102 non-diabetic Asian Indian adults with obesity in North India, an eight-week high-protein, moderately lower-carbohydrate diet produced greater weight loss, better body composition and an improved cardiometabolic profile than a standard high-carbohydrate vegetarian comparison.
• International randomised trials and meta-analyses of diets with a lower carbohydrate share report larger falls in HbA1c and liver fat over six to twelve months, with the differences narrowing as adherence wanes.

What has not been done is the study the 62% figure invites. No Indian trial has tested a sustained shift from the current carbohydrate share to something nearer the guideline band, at constant calories, over two to three years, measuring liver fat, visceral adiposity or diabetes incidence.

Four caveats sit on this evidence, and none of them is fatal to it:

• Measurement: Diet was captured by food-frequency questionnaire and 24-hour recall, methods known to under-report snacks, fats and sugars. When the invisible energy goes unreported, the visible staple takes up a larger apparent share, which can inflate the carbohydrate percentage rather than the absolute intake.
• Design: Every risk estimate quoted above comes from cross-sectional data, which cannot establish that the diet came first.
• Framing is not unanimous: Some Indian expert groups continue to support a high-carbohydrate, high-fibre pattern for type 2 diabetes, at up to 45 to 65% of energy from carbohydrate, provided most of it comes from low glycaemic index sources. The reasoning is that very low-carbohydrate patterns are hard to sustain and culturally mismatched.
• Thin intervention base: Indian trials in this area run for weeks, in samples of around a hundred people, and the long-term evidence is imported from other populations.

A figure describing 18,090 people has no opinion about any one of them. What the ICMR-INDIAB composition data supports is a structural claim:

• The average Indian energy distribution sits above the national guideline band for carbohydrate.
• It sits near the floor of it for protein.
• That pattern travels with diabetes, prediabetes and obesity in the same population.

What it does not supply is a percentage for a reader to hit, and any advice that converts a national mean into a personal target has added something the data does not contain.

The useful move is knowing what your own intake looks like and what your own glucose and lipid numbers are doing. That is a question for a registered medical practitioner or a registered dietitian working from both.

The kitchen remedies sold as a shortcut past that step are covered in our piece on popular weight-loss myths.

Know what your own numbers are doing, not the national average. A Voy clinician reads your glucose, lipid and weight picture 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.