Two Doctors Shared a BMI of 22.3, but One Carried 21 Percent Body Fat and the Other 9. A Pune Physician Has Spent Three Decades Tracing Why.

In January 2004 The Lancet printed a short letter called “The Y-Y paradox.” Its authors were Chittaranjan Yajnik, a diabetes physician at King Edward Memorial (KEM) Hospital in Pune, and John Yudkin of University College London, and the evidence was themselves. Both men had a body-mass index (BMI, weight in kilograms divided by the square of height in metres) of 22.3, a figure doctors class as healthy. Body scans, in the numbers later reported from the letter, put fat at 21.2 percent of Yajnik’s weight and 9.1 percent of Yudkin’s.

By then Yajnik’s team had been measuring pregnant women in villages near Pune, and then their children, for about ten years. The Pune Maternal Nutrition Study began in 1994, and its stated aim was to link a mother’s diet and nutritional status to her baby’s birth and to the child’s later cardiometabolic risk. What Indians eat is covered in the previous article in this series, and the chilli’s arrival in the kitchen is told in the story of Europe’s hunt for Indian spices. This one follows the body that meets the food.

Honey urine

Classical Sanskrit medical texts, more than 1,500 years old, already described the disease and named it madhumeha, honey urine. Scholars date these texts centuries apart, so no single century is safe. In a 2016 review in the World Journal of Diabetes, Marianna Karamanou and colleagues write that Sushruta, the surgeon behind the Sushruta Samhita, noted “not only the sweet taste of the urine but also its sticky feeling to the touch and its ability to attract the ants.” The review adds that he tied the disease to the rich castes and to “excessive food consumption as the rice, cereals and sweets.”

The Charaka Samhita has a chapter on prameha, a group of urinary disorders that later physicians connected with madhumeha. In the English translation on the Charaka Samhita Online site, the causes include “sedentary habits, excess sleep, curds” and “preparations of jaggery.” The same chapter sorts patients into “those who are obese and strong” and “those who are emaciated and weak.” It does not map these onto the modern types 1 and 2, and a modern reader should not do it for the text.

On heredity, the translators render the text as saying that patients “borne of prameha parents (hereditary) are not curable because of the morbidity in their bija.” Bija is the word for seed, and the translators gloss it as genes. Charaka had no concept of genes in the modern sense. What he recorded is that the disease ran in families.

A disease of the urban middle class

In British India the disease was seen mostly in one social class. David Arnold’s 2009 article in Social History of Medicine opens with the verdict: “A disease predominantly of India’s urban middle class and increasingly common in modern India, diabetes attracted little state medical attention either before or in the decades immediately following Indian independence in 1947.” Arnold describes doctors explaining it through class, race, climate and the strains of colonial modernity. We did not retrieve the nineteenth-century studies he draws on, so this article does not summarise them.

Same BMI, different body

The Y-Y paradox has since been tested on large numbers. In 2021 Rishi Caleyachetty and colleagues published, in The Lancet Diabetes and Endocrinology, a cohort of 1,472,819 people in England. They asked what BMI carries the same type 2 diabetes risk as a BMI of 30 in white adults. For South Asian adults the answer was 23.9. It was 28.1 for Black adults, 26.9 for Chinese adults and 26.6 for Arab adults.

These are figures from England, and they describe South Asians living there. They agree with a 2015 review by Naveed Sattar and Jason Gill, who write that South Asians “typically develop the disease 5-10 years earlier and at a lower BMI.” Sattar and Gill point to “high percentage of body fat and high proportion of deep subcutaneous and visceral fat,” and to low muscle mass and low fitness as the likely contributors.

The thin-fat baby

Yajnik’s group went looking for the difference at birth. In the Pune Maternal Nutrition Study’s first report, published in the International Journal of Obesity in 2003, they measured 631 term babies born in six villages near Pune and compared them with 338 born at Princess Anne Hospital in Southampton. The Indian mothers averaged 44.6 kilograms and a BMI of 18.2, against 63.6 kilograms and 23.4 for the Southampton mothers. The Indian babies weighed 2.7 kilograms on average at birth, the English babies 3.5.

The babies were smaller in every measurement, but not by the same amount. Abdominal circumference ran 2.38 standard deviations below the Southampton average. The skinfold under the shoulder blade ran only 0.53 below, the most preserved measurement of all. The authors concluded: “Small Indian babies have small abdominal viscera and low muscle mass, but preserve body fat during their intrauterine development. This body composition may persist postnatally and predispose to an insulin-resistant state.”

A second Pune report, in Diabetologia in 2008, followed 700 pregnant women and then their children to the age of 6. Two-thirds of the mothers had low vitamin B12, below 150 picomoles per litre. “The offspring of mothers with a combination of high folate and low vitamin B12 concentrations were the most insulin resistant,” the authors wrote. The design paper for a later Pune trial states that B12 deficiency “was widespread in this population, due to low dietary intake.”

The thin-fat newborn is disputed. In 2020 a Bengaluru team led by Rebecca Kuriyan measured fat directly in 156 healthy term babies of middle-income mothers in the Journal of Nutrition. Mean body fat was 9.8 percent, which they found comparable to published estimates from elsewhere. They reported that “Indian babies with normal birth weight did not demonstrate the thin but fat phenotype.” The study measured babies of normal weight in one city, and Yajnik’s finding concerned small babies in villages, so the two do not strictly contradict each other. Neither settles whether the phenotype is present at birth or develops in infancy.

What famine leaves behind

The theory that connects these findings dates to 1992, when Nicholas Hales and David Barker proposed the “thrifty phenotype.” A fetus short of food, they argued, is programmed to conserve energy, and that programming turns harmful if food later becomes plentiful. Yajnik and Krishnaveni describe the field that grew from this, the developmental origins of health and disease, as the proposal that nutrition and other conditions at critical stages of fetal growth can predispose a person to diabetes in later life.

Famines elsewhere gave the idea its human tests. In a 1998 Lancet paper, Ravelli and colleagues studied 702 people born in Amsterdam around the Dutch famine of 1944 to 1945. Those exposed in the womb had higher blood glucose two hours after a standard glucose drink. The authors concluded that “poor nutrition in utero may lead to permanent changes in insulin-glucose metabolism, even if the effect on fetal growth is small.”

The Chinese famine of 1959 to 1961 is the closest parallel to India’s history of scarcity followed by plenty. Li and colleagues at China’s National Institute for Nutrition and Food Safety, writing in Diabetes in 2010, studied 7,874 rural adults. In severely hit areas, people exposed to the famine as fetuses had 3.92 times the odds of hyperglycaemia of unexposed people. Among those who later ate an affluent, Western-style diet, the odds rose to 7.63. “This association appears to be exacerbated by a nutritionally rich environment in later life,” the authors wrote.

In Matlab, Bangladesh, Sarah Finer and colleagues reported in BMJ Open in 2016 that underweight young adults exposed to the 1974 to 1975 famine in gestation had raised blood glucose after a glucose test. All of these are associations in observational cohorts, and none is a controlled experiment.

India’s own famines have a thinner record. A 2026 study by Sankar Mukhopadhyay in Social Science and Medicine used national ageing-survey data and found that women exposed to the Bengal famine of 1943 were 3 to 4 centimetres shorter, with a height deficit of about 4.2 centimetres for those exposed in the womb. It also found worse self-rated health. It reports no diabetes outcome, and we found no published study that measures diabetes in people exposed to that famine.

Yajnik’s argument

Yajnik has made the historical case himself. In a 2024 paper in The Lancet Regional Health Southeast Asia, he notes that over the two centuries from 1830 to 1980 “Europeans gained up to 15 cm in height while Indians and SE Asians did not gain at all or lost height.” He continues: “This has been partly ascribed to colonial occupation and famines which were a result of exploitative policies. These have contributed to the evolution of the thin-fat phenotype, which has a lower threshold for cardiometabolic disease.” The historical record behind that sentence includes the famine of 1876 to 1878 and the wider economic drain.

Yajnik’s verb in that passage is contributed. In the same paper he writes: “The biological unit that transmits the DBM [double burden of malnutrition] across generations is the mother-baby dyad.” In 2017 he and G. V. Krishnaveni described the current diabetes epidemic as arising “on the backdrop of multigenerational maternal undernutrition in the country.”

Animal evidence from his lab points the same way. Anand Hardikar, Yajnik and colleagues reported in Cell Metabolism in 2015 on Wistar rats kept on a restricted diet for 50 generations. The rats had low birth weight, high visceral fat and insulin resistance, and an eight-fold greater susceptibility to diabetes triggered by streptozotocin, a chemical that damages insulin-producing cells. The abnormalities were “not reversed after two generations of unrestricted access to commercial chow.” These were rats, and the authors suggest only that the same changes may contribute in undernourished human populations.

Older than the Raj

One finding cuts against the strongest form of the colonial argument. In 2019 Emma Pomeroy, Veena Mushrif-Tripathy of Pune’s Deccan College, and colleagues published in Scientific Reports an analysis of 197 South Asian skeletons spanning 11,000 years, compared with 2,003 skeletons from around the world. South Asian skeletons indicated low lean mass throughout. “Low lean mass has characterised South Asians since at least the early Holocene,” the authors concluded, and it “may represent long-term climatic adaptation or neutral variation.”

The authors do not wave the famines away. They accept that the repeated famines of the nineteenth and early twentieth centuries “were exacerbated by British colonial policy,” and that these might have selected for genes tied to low lean mass or reduced lean mass across generations. But “given the antiquity of this phenotype,” they write, “the most likely explanations for low lean mass in South Asia appears to be either climatic adaptation or neutral evolutionary processes.” They add that the phenotype is “unlikely to change extensively in the short term.”

Genetics has not filled the gap. In 2011 Jaspal Kooner and colleagues found six new gene regions linked to type 2 diabetes in people of South Asian ancestry. Yet a 2014 review by Sohani and colleagues concluded that “there is no strong evidence to indicate that South Asians possess a greater genetic risk of type 2 diabetes than white Europeans.” Sattar and Gill likewise find “no clear evidence” of a major genetic contribution.

The popular “thrifty gene” idea, that famines selected for fat-storing genes, fares poorly in a 2016 test by Wang and John Speakman. They found signs of positive selection at 9 of 115 body-mass-index gene variants, and at 5 of those 9 the favoured variant was the one for leanness.

Read together, the two bodies of work leave this position. The lean, fat-prone build is at least 11,000 years old, famine may have added to it, and the diet it now meets, described in the earlier article, changed far more recently.

Feeding mothers is not enough

If undernourished mothers pass risk to their children, feeding the mothers should lower it. The Hyderabad Nutrition Trial tested this. From 1987 to 1990, in 29 villages near Hyderabad, pregnant and nursing women and their children under six in 15 villages received a daily protein and calorie supplement, and 14 villages did not. Allocation was not randomised. Sanjay Kinra and colleagues examined 1,826 of the children, 70 percent of the cohort, in 2009 to 2012, at an average age of 21.6, and published the results in PLOS Medicine in 2020.

The supplements “were not associated with lower levels of cardiovascular risk factors among offspring when they were young adults.” The measures included blood pressure, cholesterol, artery thickness and fasting insulin. The authors added that policymakers “should attach limited value to cardiovascular health benefits of maternal and child protein-calorie food supplementation programmes.” The authors list the limits of the study, among them the lack of randomisation and possible bias from the 30 percent of the cohort they could not follow.

The Pune team is testing a different approach. The PRIYA trial began in September 2012 and gives adolescent boys and girls vitamin B12, vitamin B12 with other micronutrients and milk powder, or a placebo, daily for at least three years or until the birth of their first child. Its design paper states the hypothesis that this “will improve birth weight, B12 status, and reduce future diabetes risk, in their offspring.” We found no published result on diabetes risk in those children.

The next generation

The numbers on adults come from the Indian Council of Medical Research’s INDIAB survey. Its 2023 report in The Lancet Diabetes and Endocrinology covered 113,043 people in 31 states and union territories, surveyed between October 2008 and December 2020. It found diabetes in 11.4 percent of adults and prediabetes in 15.3 percent. The authors wrote that the epidemic “is stabilising in the more developed states of the country” but “still increasing in most other states.” The International Diabetes Federation’s 2025 atlas puts India second to China in the number of adults with diabetes, so the phrase “diabetes capital of the world” does not hold up.

The young are being counted too. In surveys in 2006 and 2016 in southern India, Arun Nanditha and colleagues found that type 2 diabetes among people aged 20 to 39 rose from 4.5 percent to 7.8 percent. They name obesity and a family history of diabetes as the main contributors. Type 1 diabetes is a different disease that the studies above do not address. The IDF counts 301,000 Indians under 20 living with it in 2024.

Gestational diabetes, which first shows up in pregnancy, is the point where the cycle runs forward. A 2024 meta-analysis of 110 Indian articles by Mantri and colleagues put the pooled prevalence at 13 percent, with a 95 percent confidence interval of 9 to 16. The INDIAB team’s 2025 national analysis of 1,032 pregnant women, using British NICE criteria, found 22.4 percent. It found 32.9 percent in Central India and 16 percent in West India. Mantri’s team attributes part of the spread to the lack of consensus on screening and diagnosis.

A Mysore cohort shows what that exposure does. G. V. Krishnaveni and colleagues followed 514 children, 35 of them born to diabetic mothers. At age 9.5 the daughters of diabetic mothers had larger skinfolds and higher insulin resistance than children of non-diabetic parents, and sons had higher insulin resistance. The authors wrote that “the intrauterine environment experienced by ODMs [offspring of diabetic mothers] increases diabetes and cardiovascular risk over genetic factors; the effects strengthen during childhood.” The sample was small, and the finding comes from one city. Undernutrition may have passed risk down in earlier generations, and in this cohort maternal diabetes appears to pass it down now.

What it costs a family

In the Indian CKD Registry’s first report in 2012, covering 52,273 adult patients at nephrology centres, diabetic nephropathy was the commonest cause of chronic kidney disease at 31 percent. Those are patients who reached a specialist, so the figure describes who is being treated and not the whole population.

Nanda and Sharma analysed a national sample survey of health spending and reported in Health Care Science in 2023 that nearly 38 percent of households with a diabetic member had catastrophic health expenditure, by their 10 percent threshold. About 10 percent of such households were pushed below the poverty line by out-of-pocket costs. For hospital stays, 48.5 percent of households paid with distressed sources of finance.