Two people walk into a clinic at the same height and the same weight. Identical BMI, same age, same sex. One of them is metabolically healthy. The other has type 2 diabetes, a liver marbled with fat, and triglycerides high enough to threaten his pancreas. The scale cannot tell them apart. Neither can the number calculated from it.

This is not a rare quirk to be filed away as an exception. It is the central fact of body composition, and almost every measurement we actually use is blind to it. What separates those two people is not how much fat they carry. It is whether the fat is where it belongs, because each of us has a private limit for how much can be stored safely, and the two people above are sitting on opposite sides of theirs.

Whatever is actually driving metabolic disease, then, it is not mass on a scale. Which makes it worth asking where the number we judge everyone by came from, and why we ever let it speak for the person it is measuring.

The number that was never about you

The number most of us are judged by has a stranger history than its authority suggests.

In 1832 Adolphe Quetelet, a Belgian astronomer and statistician, published the ratio of weight to height squared. He was not studying obesity, and he was not treating patients. He was building what he called social physics, an attempt to describe l’homme moyen, the average man, by finding the statistical regularities in human populations. The index was a tool for characterizing the distribution of a group. It was never proposed as a diagnosis of an individual, because that was not the question he was asking.^[3]

It sat mostly unused for well over a century. Then in 1972 Ancel Keys, the physiologist, evaluated the available measures of relative weight, concluded that Quetelet’s ratio performed best of a mediocre set, and renamed it the Body Mass Index. His endorsement was explicitly for population studies. He was choosing the least bad instrument for epidemiology, not certifying a personal verdict.^[3]

Insurers and clinics adopted it anyway, because it is free, requires no equipment, and produces a tidy category. So a ratio devised by an astronomer to describe crowds, promoted by a physiologist for use on crowds, now appears on individual medical records as if it described the person sitting in the chair.

The failure is not philosophical, it is arithmetical. Weight over height squared cannot distinguish muscle from fat, so a heavily trained athlete is routinely classified as obese. It cannot see where fat is stored, so two people at an identical BMI can have entirely different amounts of fat around their organs. And it says nothing about the person who is slim on the outside and packed with visceral and liver fat on the inside, a phenotype common enough to have earned a nickname, thin outside, fat inside. That person passes the screening and has the disease.

Even the guidelines have started to concede the point. Britain’s NICE now tells clinicians to interpret BMI with caution in people with high muscle mass and in adults over 65, and recommends measuring waist against height as well.^[4] That is an official acknowledgement that the number cannot do the job alone. To see what it is missing, you have to look at what fat tissue actually is.

Fat is an organ, not a warehouse

The intuitive model of body fat is a storage depot: inert, passive, a bag of surplus calories that sits there looking bad and straining your knees. That model is wrong in a way that matters clinically.

Adipose tissue is an endocrine organ, and by mass one of the largest in the body. It secretes leptin, which reports energy availability to the brain and regulates appetite and reproductive function. It secretes adiponectin, which improves insulin sensitivity and is, counterintuitively, lower in people carrying more fat. It releases inflammatory signaling molecules including interleukin 6 and tumour necrosis factor alpha. It expresses aromatase, the enzyme that converts androgens into oestrogens, which is why fat mass changes hormone levels rather than merely accompanying them.

The clearest proof that fat is an organ comes from the rare people born almost entirely without it. In congenital generalized lipodystrophy, an inherited condition affecting a handful of people in a million, there is almost no adipose tissue from birth, and the result is not metabolic health but its collapse: severe insulin resistance, aggressive early diabetes, dangerous triglycerides, a liver saturated with fat. With nowhere to store incoming lipid, it lands in the liver and muscle instead.^[1] These patients are also profoundly short of leptin, the hormone fat is supposed to secrete, and giving it back reverses much of the damage.^[2] It is an extreme almost nobody will encounter, and the point is emphatically not that fat is harmless, which the rest of this essay should dispel. What the condition isolates is the principle: you can be gravely sick from fat in the wrong place even when there is almost none of it, because what has failed is an organ, not a quantity on a scale.

The second correction is harder to accept, because it inverts the usual moral framing. Subcutaneous fat, the layer under the skin that people dislike in the mirror, is not the pathology. It is the safe storage compartment. It is where your body is supposed to put surplus energy, sequestered away from the tissues that will be damaged by it. Well-functioning subcutaneous fat is metabolically protective.

Which reframes the whole question. The problem is not that you have a parking lot. The problem begins when the parking lot is full and cars start parking on the lawn.

Your personal fat threshold

That is close to the literal mechanism, and it has a name.

Roy Taylor, working at Newcastle, proposed the twin cycle hypothesis and then the personal fat threshold. The idea is that each individual has their own limit for how much fat can be stored safely in subcutaneous tissue, and that this limit is set largely by genetics and varies enormously between people. Below your threshold, the system copes. Above it, the surplus has nowhere legitimate to go, so it accumulates ectopically, in the liver, the pancreas, the muscle, and around the heart, where it interferes with the function of those organs directly.^[5]

This single idea explains the clinical observations that BMI cannot. It explains why one person develops type 2 diabetes at a BMI of 23 while another remains metabolically normal at 35: they have different thresholds, and only one of them has exceeded their own. It explains why South Asian populations develop metabolic disease at substantially lower body weights, having on average less subcutaneous storage capacity. It explains why losing a relatively modest amount of weight can produce diabetes remission, because you do not need to become slim, you only need to drop back under your own threshold and let the liver and pancreas clear.

The molecular consequences of that spillover, how lipid inside a muscle or liver cell actually breaks insulin signaling, I have covered in detail in the piece on type 2 diabetes, and I will not repeat it here. What matters for this essay is the accounting. Fat itself is not the enemy. Fat in the wrong compartment is. And the wrong compartment fills only when the right one is full.

Once that spillover begins, it also starts to defend itself.

The loop that tightens

Here is where most discussions of this topic get the causality backwards, and I think it is worth being precise, because the correct version is both more useful and less moralizing.

The assumption is that people become heavier because they are sedentary. The longitudinal evidence points substantially the other way. The EarlyBird study followed children annually and tested both directions explicitly. Body fat predicted subsequent reductions in physical activity. Physical activity did not predict subsequent changes in body fat. The authors were blunt about the conclusion: inactivity appears to be the result of fatness rather than its cause.^[6] A separate longitudinal cohort in children aged eight to eleven reported the same asymmetry, fatness predicting decreased activity and increased sedentary time, and not the reverse.^[7] A Mendelian randomization analysis, which uses genetic variants allocated at conception and is therefore not vulnerable to the confounding that plagues observational work, again supported adiposity driving activity levels.^[8]

I want to be careful with this, because the strongest data here is in children and I am not going to pretend it settles the adult case. But the direction is consistent, and the mechanism is not mysterious or a question of character. Carrying additional mass raises the metabolic cost of every step. It loads the knees and hips, and osteoarthritic pain is an excellent deterrent to movement. It produces breathlessness on exertion. It drives obstructive sleep apnoea, which fragments sleep and delivers people into the next day exhausted. Chronic inflammatory signaling produces fatigue directly. None of that is a failure of willpower. It is a body making movement more expensive and less pleasant, and then getting less of it.

That is what makes this a loop rather than a state. Less movement means less muscle. Less muscle matters more than most people realize, because skeletal muscle is the largest disposal site for glucose after a meal, so losing it shrinks the buffer that was protecting you. A smaller buffer means more circulating fuel to store, which pushes you further past your threshold, which makes movement harder still. The worst version of this is sarcopenic obesity, low muscle and high fat together, which carries a worse prognosis than either alone and which BMI is completely blind to, because muscle and fat weigh the same on the scale that is judging you.

So the loop tightens quietly, and while it does, the bill accumulates in specific organs.

What it actually costs

None of what follows is speculative. These are among the best-characterized associations in medicine, and where I can point at causal evidence rather than correlation, I will.

Type 2 diabetes. The most direct consequence of the mechanism above, since the pancreas and liver are the organs where spilled fat lands first. The relationship is dose dependent and, importantly, reversible in a way most chronic disease is not.

Cardiovascular disease. Excess visceral fat produces a characteristic and dangerous lipid pattern: high triglycerides, low HDL, and a shift toward small dense LDL particles. It raises blood pressure. It promotes atrial fibrillation. And it drives a specific form of heart failure, the kind with preserved ejection fraction, where the heart pumps normally but cannot relax and fill properly. Fat accumulating directly on the heart, epicardial adipose tissue, is now understood as central to that phenotype, acting both as a local inflammatory organ and as a mechanical restraint on a heart trying to expand.^[9]

Fatty liver. Metabolic dysfunction associated steatotic liver disease, recently renamed from the older non-alcoholic terminology, is the most common chronic liver disease on Earth and affects more than 30 percent of adults.^[10] It is the liver arm of the spillover story, and it is not benign. A meaningful fraction progresses to inflammation, then fibrosis, then cirrhosis, then liver cancer, and it does all of that without symptoms until late.

Cancer. This is the association the public underestimates most severely. In 2016 an International Agency for Research on Cancer working group of 21 independent experts reviewed the evidence and concluded that absence of excess body fatness reduces the risk of thirteen separate cancers: colon and rectum, oesophageal adenocarcinoma, kidney, postmenopausal breast, endometrium, gastric cardia, liver, gallbladder, pancreas, ovary, thyroid, meningioma, and multiple myeloma.^[11] Three mechanisms carry most of it. Chronically elevated insulin and IGF-1 are growth signals, and growth signals applied continuously to cells that have acquired mutations are exactly what you do not want. Aromatase in adipose tissue raises oestrogen exposure, which drives the postmenopausal breast and endometrial cases. And chronic low-grade inflammation supplies the rest.

Dementia. Here the evidence needs care, and it is the most interesting item on the list for reasons that go beyond the disease itself. Higher body fat in midlife, roughly the forties and fifties, is associated with increased risk of later dementia. But measure the same relationship in people over 70 and it inverts, with higher BMI now appearing protective.^[12] That reversal looks like a contradiction, and it is the single best demonstration of the trap that runs through this entire literature.

The paradox that is not one

You will have seen the headlines. Overweight people survive heart failure better. Higher BMI predicts better outcomes in dialysis, in cancer, in old age. The obesity paradox is real as a statistical observation and it is repeatedly presented as evidence that the concern is overblown. It is mostly an artifact, and the ways it is generated are worth knowing, because they recur everywhere in health research.

Start with the dementia inversion, because the explanation there is clean. Dementia has a prodromal phase measured in decades, and weight loss is one of its early features, appearing years before any cognitive diagnosis. Genetic risk for Alzheimer’s is associated with accelerated weight loss beginning in the late forties, and amyloid burden predicts weight loss even in people who are still cognitively normal.^[12] So when you measure BMI in a 75 year old and follow them for dementia, you are not measuring whether fat protects the brain. You are partly measuring who has already begun losing weight because their disease started fifteen years ago. The arrow runs backwards, and the statistics faithfully report it forwards.

That is reverse causation, and it does most of the work. Smoking supplies the second bias: smokers are leaner and die more, which loads the lean group with deaths that have nothing to do with leanness. When researchers restricted analysis to people who had never smoked and accounted for reverse causation, the paradox in cardiovascular disease did not merely shrink, it reversed, and the lowest mortality returned to the BMI range of 20 to 25.^[13]

The third bias is subtler and worth naming properly. If you study only people who already have a disease, say heart failure, you have conditioned your sample on that disease. Being thin and having heart failure implies some other cause was strong enough to produce it, and that other cause carries its own mortality. So among the sick, the thin look worse, not because thinness is bad but because you selected a group in which thinness implies hidden severity. This is collider bias, and it is why studying survival within a diseased population tells you very little about what caused the disease.^[14]

Strip those three out and the picture resolves. When the question is put to Mendelian randomization, where genetic variants that raise adiposity are effectively randomized at conception and cannot be confounded by smoking, illness, or social class, higher adiposity comes out causally associated with coronary artery disease.^[15] I used the same technique in the cholesterol essay to separate what LDL actually does from what observational data merely suggested. It gives the same answer here that the mechanism predicts.

So both things are true at once, and the honest position holds both. BMI is a poor instrument for judging an individual. Excess adiposity, properly measured, is genuinely and causally harmful. People who discover the first fact often conclude the second is false, and that is a mistake with a body count.

What to measure instead

If the scale is a poor instrument and BMI is worse, the practical question is what to use in their place. The answer is not exotic and most of it costs nothing.

Measure your waist, and compare it to your height. The target is a waist under half your height, and NICE now recommends exactly this alongside BMI, because it captures central adiposity, which is the fat that matters, and works across sexes and ethnicities in a way BMI does not.^[4] A tape measure outperforms the scale here for the simple reason that it is looking at the right compartment.

If you want the real picture, a DEXA scan gives you fat mass, lean mass, and their distribution separately, which is the actual variable this essay has been about. I covered where it fits among the various measurement options in the piece on biological age tests. Once a year is plenty, and the trend matters more than any single reading.

Then look in the blood, because spillover is visible there long before it is visible anywhere else. The triglyceride to HDL ratio, HbA1c, fasting insulin, and ALT together tell you whether fat is arriving in places it should not be, and they move years before a diagnosis does. The blood tests series covers what each one is actually reporting.

And track your muscle, not just your fat, because it is half of body composition and the half that is protective. Grip strength and what you can lift are usable proxies, and resistance training is the intervention that moves them.

None of these ask what you weigh. They ask where it is, what it is made of, and whether your storage is holding. Those are the questions the biology is actually answering.

The scale has been measuring the one thing that matters least, and hiding the two that matter most.

---

1. Akinci, B., Meral, R., & Oral, E. A. (2020). Congenital generalized lipodystrophies: new insights into metabolic dysfunction. https://pmc.ncbi.nlm.nih.gov/articles/PMC7605893/

2. Frontiers in Endocrinology. (2026). Twenty years of metreleptin therapy in congenital generalized lipodystrophy type 1: the longest reported follow-up to date. https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2026.1815903/full

3. Eknoyan, G. (2008). Adolphe Quetelet (1796-1874), the average man and indices of obesity. Nephrology Dialysis Transplantation, 23(1), 47–51. https://academic.oup.com/ndt/article/23/1/47/1923176

4. National Institute for Health and Care Excellence. Identifying and assessing overweight, obesity and central adiposity (NG246). https://www.nice.org.uk/guidance/ng246/chapter/Identifying-and-assessing-overweight-obesity-and-central-adiposity

5. Taylor, R., & Holman, R. R. Pathogenesis and remission of type 2 diabetes: what has the twin cycle hypothesis taught us? https://pmc.ncbi.nlm.nih.gov/articles/PMC7673778/

6. Metcalf, B. S., Hosking, J., Jeffery, A. N., Voss, L. D., Henley, W., & Wilkin, T. J. (2011). Fatness leads to inactivity, but inactivity does not lead to fatness: a longitudinal study in children (EarlyBird 45). Archives of Disease in Childhood, 96(10), 942–947. https://pubmed.ncbi.nlm.nih.gov/20573741/

7. Hjorth, M. F., Chaput, J.-P., Ritz, C., Dalskov, S.-M., Andersen, R., Astrup, A., et al. (2014). Fatness predicts decreased physical activity and increased sedentary time, but not vice versa: support from a longitudinal study in 8- to 11-year-old children. International Journal of Obesity, 38, 959–965. https://www.nature.com/articles/ijo2013229

8. Richmond, R. C., Davey Smith, G., Ness, A. R., den Hoed, M., McMahon, G., & Timpson, N. J. (2014). Assessing causality in the association between child adiposity and physical activity levels: a Mendelian randomization analysis. PLOS Medicine, 11(3), e1001618. https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1001618

9. van Woerden, G., van Veldhuisen, D. J., Gorter, T. M., et al. (2021). Epicardial fat expansion in diabetic and obese patients with heart failure and preserved ejection fraction, a specific HFpEF phenotype. https://pmc.ncbi.nlm.nih.gov/articles/PMC8484763/

10. Current status and future trends of the global burden of MASLD. Trends in Endocrinology and Metabolism. https://www.sciencedirect.com/science/article/abs/pii/S1043276024000365

11. Lauby-Secretan, B., Scoccianti, C., Loomis, D., Grosse, Y., Bianchini, F., & Straif, K. (2016). Body fatness and cancer, viewpoint of the IARC Working Group. New England Journal of Medicine, 375(8), 794–798. https://www.nejm.org/doi/full/10.1056/NEJMsr1606602

12. Li, J., et al. Association of genetic risk score for Alzheimer’s disease with late-life body mass index: evaluating reverse causation. https://pmc.ncbi.nlm.nih.gov/articles/PMC11972977/ · Mid- to late-life body mass index and dementia risk: 38 years of follow-up of the Framingham study. https://pmc.ncbi.nlm.nih.gov/articles/PMC8796797/

13. Stokes, A., & Preston, S. H. (2015). Smoking and reverse causation create an obesity paradox in cardiovascular disease. Obesity, 23(12), 2485–2490. https://pubmed.ncbi.nlm.nih.gov/26421898/

14. Banack, H. R., & Kaufman, J. S. (2014). Obesity paradox: conditioning on disease enhances biases in estimating the mortality risks of obesity. Epidemiology, 25(3), 379–380. https://journals.lww.com/epidem/Fulltext/2014/05000/Obesity_Paradox__Conditioning_on_Disease_Enhances.17.aspx

15. Assessing causal estimates of the association of obesity-related traits with coronary artery disease using a Mendelian randomization approach. https://pmc.ncbi.nlm.nih.gov/articles/PMC5940685/