Which diet

Walk past the diet shelf in any bookshop and you'll see a dozen books each sure its own plan is the one: cut the carbohydrate, cut the fat, eat in a window, eat like your ancestors. They can't all be best. The way to find out is to take a few hundred people, put them on different diets at random, and weigh them a year later. That has been done many times, and the answer is more interesting than any of the books. This lesson reads those trials. Like lesson 3, it describes what studies found about weight and says nothing about what yours should be.

Low-fat against low-carbohydrate, for a year

The cleanest test of the two most argued-over diets is a trial called DIETFITS, published by Gardner and colleagues in 2018.¹ It took 609 adults aged 18 to 50, with a BMI between 28 and 40 and no diabetes, and randomised them to a healthy low-fat diet or a healthy low-carbohydrate diet for twelve months. Both groups went to 22 small-group sessions with health educators. In the words of the abstract, which is what this course read, the sessions "focused on ways to achieve the lowest fat or carbohydrate intake that could be maintained long-term and emphasized diet quality."

Notice what "healthy" is doing in both names. Neither group was told to eat whatever it liked within a macronutrient limit. Both were steered towards better food, and the only planned difference was which of fat or carbohydrate they cut.

The trial had a second question, and it's the one that makes DIETFITS famous. The abstract opens by saying that "Previous research suggests genotype or insulin-glucose dynamics may modify the effects of diets." So the investigators measured two things at the start that might predict who does better on which diet: a genetic pattern built from three gene variants, and how much insulin each person released 30 minutes after a dose of glucose. In this course's reading, the insulin measure tests a prediction of lesson 3's carbohydrate-insulin model: people who secrete more insulin should do better cutting carbohydrate.

Two details keep this from proving more than it does. First, what people actually ate. At twelve months the low-fat arm was eating 48 percent of energy from carbohydrate and 29 percent from fat, and the low-carbohydrate arm 30 percent and 45 percent.¹ Thirty percent carbohydrate is a long way from a ketogenic diet, so DIETFITS tells you nothing directly about eating very little carbohydrate. The evidence this course read on eating that little comes from people with type 2 diabetes, a trial run by the diet's proponents that wasn't randomised and pooled trials that were, and lesson 9 takes it up. Second, 481 of the 609 finished (79 percent), and a year is a year, not five.

Run lesson 1's question over it. The low-fat arm ate carbohydrate instead of fat, the low-carbohydrate arm ate fat instead of carbohydrate, and both were coached towards less refined food. Protein was nearly the same in both arms: 21 against 23 percent. So the trial compared two swaps inside a shared change, and the shared change may be most of why both groups lost weight. That is this course's reading, not the authors'.

Four named diets, and who stuck with them

Thirteen years before DIETFITS, Dansinger and colleagues in Boston put four famous diets side by side: Atkins (cutting carbohydrate), Zone (balancing the macronutrients), Weight Watchers (cutting calories) and Ornish (cutting fat).² They randomised 160 adults, 40 to each diet. Everyone had a BMI between 27 and 42 and high blood pressure, raised cholesterol or raised fasting blood sugar. This course read the abstract, and it describes a twist in the design: "After 2 months of maximum effort, participants selected their own levels of dietary adherence."

At a year, with anyone who dropped out counted as having lost nothing, the average losses were:

The authors' summary: "Each popular diet modestly reduced body weight and several cardiac risk factors at 1 year." The two diets that asked for the biggest change from ordinary eating, Atkins and Ornish, were the two that lost the most people.

The central finding is one sentence: "Amount of weight loss was associated with self-reported dietary adherence level (r = 0.60; P<.001) but not with diet type (r = 0.07; P = .40)."² The r is a correlation, as in lessons 2 and 3: 0 means no relationship, and 1 means two things move in perfect step. There's no fixed bar a correlation has to pass to count. In this course's reading, 0.60 is a strong relationship for anything as messy as people and their weight, and 0.07 is next to none. How well people stuck to their diet went with how much they lost. Which diet they'd been given barely did. And sticking was hard for everyone: "Overall dietary adherence rates were low".

The average and the spread

Look again at the standard deviation column. (DIETFITS's abstract gives no measure of spread, so the picture comes from Dansinger.) A standard deviation measures how far individuals typically sit from the average. The biggest gap between any two diets' averages is 1.2 kg, Ornish against Atkins. The standard deviations run from 4.8 to 7.3 kg, four to six times that gap. (This course's arithmetic: 3.3 minus 2.1 is 1.2; 4.8 divided by 1.2 is 4, and 7.3 divided by 1.2 is about 6.)

So the differences between people inside one diet dwarf the differences between diets. Take Ornish. One standard deviation either side of its 3.3 kg average runs from a gain of 4.0 kg to a loss of 10.6 kg. For a bell-shaped spread, about two thirds of people fall inside that band and a third outside it. That rule of thumb is this course's gloss, and the abstract does not say the spread was bell-shaped; counting every dropout as zero change will have piled people up at zero.

The point survives the caveat. Within the same diet, some people lost a lot, some lost nothing and some gained.

Here's the idea people most often get wrong. A wide spread looks as if it proves each person has a diet that suits them. It does not. It proves people differ in how much they lose, which could come from how well they stuck to it, their circumstances, illness, measurement, chance, or a real match between person and diet. A parallel trial cannot separate these, because each person tried one diet. DIETFITS went looking for the match with the two best candidates it had, and didn't find it.

Pooling the named diets

In 2014 Johnston and colleagues pooled 48 randomised trials of named diets, 7,286 people, in a network meta-analysis, the method Mental Fitness lesson 2 read.³ This course read the abstract, which says, "Our analyses adjusted for behavioral support and exercise."

Compared with no diet, at twelve months, low-carbohydrate diets came to 7.25 kg of weight loss (95% credible interval 5.33 to 9.25) and low-fat diets 7.27 kg (5.26 to 9.34). A credible interval is the Bayesian counterpart of a confidence interval, and you can read it the same way here; that gloss is this course's. "Weight loss differences between individual diets were minimal." The authors' conclusion: "This supports the practice of recommending any diet that a patient will adhere to in order to lose weight."³

Two cautions. These are losses against eating no particular diet, not against each other, and they are 6 and 12 month results, silent about five years. This course adds one reading of its own: near-identical figures are what you'd expect if both diets work mainly by getting people to eat less. A carbohydrate-insulin proponent would reply that a low-carbohydrate diet gets people to eat less by a different route, and that is lesson 3's dispute, which these figures do not settle.

One more pair of figures in the abstract is worth a look. Behavioural support was associated with 3.23 kg more loss at six months, more than the 1.71 kg the abstract gives as its example of a gap between two named diets (Atkins over Zone). By twelve months the support estimate was 1.08 kg, with an interval from -1.82 to 3.96 that crosses zero, while exercise went the other way, from 0.64 kg to 2.13 kg (0.43 to 3.85). These come from comparing across trials, not from randomising people to support or exercise, so they're associations. This course's reading: what surrounds a diet can matter as much as its name, and which part matters seems to change with time.

England's obesity guideline, NICE NG246, updated in January 2026, reaches a compatible position in its own terms.⁷ Dietary approaches should "keep the person's total energy intake below their energy expenditure", and "This could be done by lowering specific macronutrient content (for example, low-fat or low-carbohydrate diets) or using other methods to limit overall energy intake." It adds a warning: "Advise people not to use restrictive diets that are nutritionally unbalanced, because they are ineffective in the long term and can be harmful." This course read the adult sections of the assessment chapter and the dietary approaches section in full.

Eating in a window

Time-restricted eating means eating only within a set number of hours each day, commonly eight. The claim you'll most often meet in diet books is that the window itself does something: that fasting for sixteen hours switches the body to burning fat, so you lose more than the calories would predict. This course has not read a proponent's own statement of it, so take that as the popular version. Two trials tested it in different ways.

TREAT, by Lowe and colleagues and published in 2020, ran for twelve weeks through a phone app with 116 adults.⁴ One group ate only between noon and 8 pm; the other ate three structured meals a day. Neither had a calorie target. The difference in weight between groups was -0.26 kg, with a 95 percent confidence interval from -1.30 to 0.78 (P = .63), and in the 50 people tested in person, "There were no differences in estimated energy intake between groups." The conclusion: "Time-restricted eating, in the absence of other interventions, is not more effective in weight loss than eating throughout the day." This course read the abstract.

The abstract also reports lower lean mass in the arms and legs in the window group, in that same subset of 50. It isn't what the trial was designed to test, and the subgroup is small, so treat it as a flag to watch for in bigger trials, not a finding.

TREAT asked whether a window helps when nothing else changes. The next trial asked a sharper question.

Same calories, different clock

Liu and colleagues in China randomised 139 adults with obesity for a full year.⁵ Both groups were put on the same calorie-restricted diet, with the allowance set separately for men and women. One group also ate only between 8 am and 4 pm, early in the day, where TREAT's window ran into the evening, so the two trials didn't test the same clock. The other group ate its calories whenever it chose. 118 people (84.9 percent) finished. This course read the abstract.

The answer at a year: "Changes in weight were not significantly different in the two groups at the 12-month assessment (net difference, -1.8 kg; 95% CI, -4.0 to 0.4; P = 0.11)." Body fat, waist, blood pressure and the metabolic measures went the same way.⁵

Two things to notice about that result. The net difference, 1.8 kg, is not the simple gap between the two averages, which is 1.7 kg (this course's arithmetic, 8.0 minus 6.3); the paper's figure comes from its own analysis, which the abstract does not describe. And the interval runs from a 4.0 kg advantage for the window to a 0.4 kg advantage for the other group. So the trial found no detectable extra benefit, and it does not rule out a modest one. What it makes unlikely, in this course's reading, is a window effect much bigger than 4 kg a year. An extra kilogram or two from timing alone is still inside the interval, so the trial does not settle the fat-burning claim. What it does show is that most of what this time-restricted diet achieved, the same calories eaten across the day achieved too.

A 2025 review in the BMJ by Semnani-Azad and colleagues pooled 99 trials and 6,582 adults, most of them with an existing health condition.⁶ "All intermittent fasting and continuous energy restriction diet strategies reduced body weight when compared with ad-libitum diet", meaning eating freely. Against ordinary calorie restriction, only alternate-day fasting showed an advantage, 1.29 kg (95 percent confidence interval 0.59 to 1.99, moderate certainty), and in the 17 trials of 24 weeks or longer the only differences were against eating freely. The conclusion is carefully hedged: "The current evidence provides some indication that intermittent fasting diets have similar benefits to continuous energy restriction for weight loss and cardiometabolic risk factors. Longer duration trials are needed to further substantiate these findings." This course read the abstract. NICE NG246 makes no practice recommendation on intermittent fasting and points to a research recommendation instead.⁷

After the first year

Losing weight for a year is one question. Keeping it off is another, and the popular answer is that almost nobody does.

One thing lesson 3 set out applies to every page of this section. An eating disorder can occur at any body size, and in teenage cohorts dieting went with later eating disorders, an association from observational cohorts rather than proof that the dieting caused them. If reading about weight loss stirs up worries about your own eating, the helplines in the callout at the top are there for that.

Five years after a programme

Anderson and colleagues pooled 29 US studies of structured weight-loss programmes with at least two years of follow-up, in 2001.⁸ They open with the belief they were testing: "Current perception is that participants of a structured weight-loss program regain all of their weight loss within 5 y." This course read the abstract.

At four to five years, people who had completed very-low-energy diets had kept off an average of 7.1 kg, 29 percent of what they'd lost, and people who had completed ordinary reduced-calorie diets 2.0 kg, 17 percent. The conclusion: "Five years after completing structured weight-loss programs, the average individual maintained a weight loss of >3 kg and a reduced weight of >3% of initial body weight."⁸

That first figure needs a fence around it. A very-low-energy diet means under 800 kcal a day, and NICE says these should be used only within specialist services, should "last no more than 12 weeks", and should not be used "as a long-term strategy to manage obesity".⁷ The figures are also for people who completed their programmes; the abstract calls the diets it pooled "Successful".

Read those percentages both ways. Most of the weight came back, on average: 71 to 83 percent of it (this course's arithmetic, 100 minus 29 and 100 minus 17). But not all of it, which is what the popular belief says. Those are averages, though. An average of 2 kg kept off fits most people regaining everything while a few keep off a lot, just as well as it fits everyone keeping a little. It can't tell you how many people kept weight off; Look AHEAD's eight-year figures, below, come closer. And only about 55 and 80 percent of people were followed up at four to five years, so those who came back to be weighed may not be like those who didn't. The studies are old, and all American.

The longest trial

Look AHEAD is one of the largest and longest lifestyle trials. It randomised 5,145 adults with type 2 diabetes and overweight or obesity, at 16 US centres, to an intensive programme of eating less and moving more, or to "diabetes support and education", which meant periodic group education only.⁹¹⁰ Its question wasn't weight. It was whether losing weight would cut cardiovascular deaths, heart attacks, strokes and hospital stays for angina. This course read the abstracts of both papers.

Weight loss was greater in the intervention group throughout: "8.6% vs. 0.7% at 1 year; 6.0% vs. 3.5% at study end". At eight years, "50.3% and 35.7%, respectively, lost ≥5%" of their starting weight.¹⁰

Both halves of Look AHEAD belong in an honest account. Sustained, modest loss was possible for about half of people with intensive support; the authors call it "clinically meaningful weight loss (≥5%) at year 8 in 50% of patients".¹⁰ The intervention also brought "greater reductions in glycated hemoglobin and greater initial improvements in fitness and all cardiovascular risk factors, except for low-density-lipoprotein cholesterol levels."⁹ Glycated hemoglobin is a measure of average blood sugar over the past few months, and LDL cholesterol is lesson 6's subject. And yet the events it was designed to prevent didn't fall over about ten years. That's one outcome, in people with type 2 diabetes, and it does not tell you about other outcomes or other people.

NICE's guideline tells clinicians to say something the evidence above supports: to discuss "that weight regain is likely to happen, and if it does it is not because they or their healthcare professional have 'failed'". It also asks them to "Encourage people to improve their dietary intake even if this does not result in them losing weight, because there can be other health benefits".⁷

A register of people who kept weight off

If you've read about weight maintenance, you've probably met the National Weight Control Registry, a register of people who lost a lot of weight and kept it off. Wing and Phelan's 2005 review describes it.¹¹ Members "have lost an average of 33 kg and maintained the loss for more than 5 y". They report a lot of physical activity, about an hour a day, a low-calorie and low-fat diet, eating breakfast regularly, weighing themselves often, and eating the same way on weekdays and weekends. This course read the abstract.

Memory lesson 7 gave you the question for this: how many people, and who?

Where medicine has moved

This course doesn't evaluate drugs, but it would mislead you to leave out where obesity treatment has gone. In 2021 the STEP 1 trial gave 1,961 adults without diabetes either a weekly injection of semaglutide, a GLP-1 drug, which acts on appetite, or a placebo, both with a lifestyle programme, for 68 weeks. Weight changed by "-14.9% in the semaglutide group as compared with -2.4% with placebo".¹² The trial was funded by Novo Nordisk, which makes the drug. In an exploratory follow-up of 327 people for a year after everyone stopped both the drug and the lifestyle programme, "participants regained two-thirds of their prior weight loss".¹³ Both abstracts were read. The World Health Organization issued a guideline on these drugs in December 2025, which this course hasn't read.

In this course's reading, losses of that size are two to three times what the diet trials above typically achieved. Whether a drug is right for anyone is a question for a doctor.

The WHO's own fact sheet, dated 8 December 2025, describes obesity as "a chronic, relapsing disease arising from complex interactions between genetics, neurobiology, eating behaviours, access to healthy diet, market forces, and the broader environment". It then states a position, on the ground that "Dietary and physical activity patterns are largely the result of environmental and societal conditions that greatly constrain personal choice": "Obesity is a societal rather than an individual responsibility".¹⁴ The first of those mixes an empirical claim, about what obesity arises from, with a classification that others draw differently: a 2025 Lancet commission keeps the word illness for what it calls clinical obesity, and describes "preclinical obesity as a state of excess adiposity with preserved function of other tissues and organs".¹⁵ The second is a value judgement by WHO, not a finding, and the course reports it as WHO's.

What people get wrong

"The best diet is the one that suits your insulin, or your genes." DIETFITS tested both predictors and neither picked out who did better on which diet. That is two candidates in one trial of two fairly moderate diets, so it does not rule out a predictor nobody has found yet. It does mean the two best-known candidates failed the one test this course read.

"Eating in a window burns more fat." With calories matched, Liu's window group lost no significantly more weight or body fat over a year, though the interval doesn't rule out a modest extra loss. Without matched calories, TREAT's window made no difference either. A window may help some people eat less, which is a calorie effect.

"Low-carb beats low-fat" and "low-fat beats low-carb." At a year, with both diets built around food quality, DIETFITS found 0.7 kg between them, and the interval crossed zero. Johnston's pooled figures against no diet were 7.25 and 7.27 kg.

"Everyone regains everything." On average most weight came back, but not all: among people who completed programmes in Anderson's pooled studies, 2.0 kg stayed off at four to five years after ordinary diets and 7.1 kg after very-low-energy ones, and half of Look AHEAD's intervention group was at least 5 percent lighter at eight years. The stronger version of the claim, that for most people a diet alone does not keep off enough weight to matter, sits closer to the ordinary-diet figures (2.0 kg, 17 percent). This course has not read a primary statement of that position, so it reports the figures and leaves the judgement of "enough" to you.

"Regain means you didn't try hard enough." Adherence was low in every arm of Dansinger's trial, whatever the diet, and lesson 3 showed the body resisting weight loss. NICE tells clinicians that regain "is likely to happen" and isn't a failure. The trials describe how hard sustained change is for most people, not a flaw in the people who regain.

Practice

Connections

Back. Lesson 3's energy account applies to every arm of these trials: whatever each diet did, the people who lost weight ate less than they burned. The account does not say why the diets came out so close. The energy-balance model's answer is that each cut intake about equally; DIETFITS's insulin test was one prediction of the carbohydrate-insulin model's alternative, a single trial of it and not a verdict on the model. Lesson 2's warning about self-report is the first caveat on Dansinger's adherence finding, and Memory lesson 7's question about who is in a sample is the one to ask of the registry.

Forward. These trials varied carbohydrate, fat and protein and found the split mattered less than expected for weight. Lesson 5 turns to the numbers you're given for each nutrient, what an RDA is, and why the protein figure became a dispute in 2026.

Go deeper

• Gardner and colleagues, 2018, free at PubMed Central. This course read the abstract. The largest single test of low-fat against low-carbohydrate for weight in this lesson, with the insulin and genotype predictors built in.
• Dansinger and colleagues, 2005. This course read the abstract. The source of the adherence finding everybody quotes.
• Liu and colleagues, 2022 in the New England Journal of Medicine. This course read the abstract. A model of how to isolate one variable in a diet trial.
• NICE NG246, free online. This course read the adult assessment chapter and the dietary approaches section. What a national guideline makes of the trials in this lesson, including its wording about regain.

Sources

1. C. D. Gardner and colleagues, "Effect of Low-Fat vs Low-Carbohydrate Diet on 12-Month Weight Loss in Overweight Adults and the Association With Genotype Pattern or Insulin Secretion: The DIETFITS Randomized Clinical Trial", JAMA 319(7), 2018, pp. 667 to 679, doi 10.1001/jama.2018.0245. Read: the abstract. Supports: the design, the sessions, the achieved diets, the weight results, both interactions and the quotations. That insulin secretion tests a prediction of the carbohydrate-insulin model, that 30 percent carbohydrate is far from ketogenic, the gloss of a P value and the shared-change reading are this course's.
2. M. L. Dansinger and colleagues, "Comparison of the Atkins, Ornish, Weight Watchers, and Zone diets for weight loss and heart disease risk reduction: a randomized trial", JAMA 293(1), 2005, pp. 43 to 53, doi 10.1001/jama.293.1.43. Read: the abstract. Supports: the design, the table and every quotation. The completion pattern, the reading of the correlations, the two problems with the adherence correlation and the spread arithmetic are this course's.
3. B. C. Johnston and colleagues, "Comparison of weight loss among named diet programs in overweight and obese adults: a meta-analysis", JAMA 312(9), 2014, pp. 923 to 933, doi 10.1001/jama.2014.10397. Read: the abstract. Supports: the 48 trials, 7,286 people, the 12-month figures, the behavioural-support and exercise figures and the quotations. The two cautions, the reading that both diets work mainly by reducing intake, and the gloss of a credible interval are this course's.
4. D. A. Lowe and colleagues, "Effects of Time-Restricted Eating on Weight Loss and Other Metabolic Parameters in Women and Men With Overweight and Obesity: The TREAT Randomized Clinical Trial", JAMA Internal Medicine 180(11), 2020, pp. 1491 to 1499, doi 10.1001/jamainternmed.2020.4153. Read: the abstract. Supports: the design, the between-group result, the in-person subset of 50, the energy intake sentence, the lean mass result and the conclusion. Treating the lean mass result as a flag is this course's reading.
5. D. Liu and colleagues, "Calorie Restriction with or without Time-Restricted Eating in Weight Loss", New England Journal of Medicine 386(16), 2022, pp. 1495 to 1504, doi 10.1056/NEJMoa2114833. Read: the abstract. Supports: the design, the window, completion, both arms' results, the net difference and the secondary outcomes. That it doesn't rule out a modest benefit, and the 1.7 kg subtraction, are this course's.
6. Z. Semnani-Azad and colleagues, "Intermittent fasting strategies and their effects on body weight and other cardiometabolic risk factors: systematic review and network meta-analysis of randomised clinical trials", BMJ 389, 2025, e082007, doi 10.1136/bmj-2024-082007. Read: the abstract; the correction notice, which corrects one co-author's affiliation. Supports: the 99 trials, 6,582 adults, the alternate-day figure, the longer-trial result and the quotations.
7. National Institute for Health and Care Excellence, Overweight and obesity management, NICE guideline NG246, published 14 January 2025, last updated 8 January 2026, nice.org.uk. Read: the chapter on identifying and assessing overweight, obesity and central adiposity (adult sections in full), the dietary approaches section of "Physical activity and diet" in full, and the update information, 24 September 2026. Supports: recommendations 1.16.2, 1.16.3, 1.16.7 and 1.16.12 as quoted, the definition of very-low-energy diets and the limits on their use as quoted, and the absence of a practice recommendation on intermittent fasting.
8. J. W. Anderson and colleagues, "Long-term weight-loss maintenance: a meta-analysis of US studies", American Journal of Clinical Nutrition 74(5), 2001, pp. 579 to 584, doi 10.1093/ajcn/74.5.579. Read: the abstract. Supports: the 29 studies, that the figures are for people who completed programmes, the 4 to 5 year figures, the follow-up rates and the quotations. The percentages regained and the follow-up caveat are this course's.
9. Look AHEAD Research Group, "Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes", New England Journal of Medicine 369(2), 2013, pp. 145 to 154, doi 10.1056/NEJMoa1212914. Read: the abstract. Supports: the design, the weight figures, the early stop, the event counts and rates, the hazard ratio, the risk-factor sentence and the conclusion. The glosses of futility, person-years and glycated hemoglobin are this course's.
10. Look AHEAD Research Group, "Eight-year weight losses with an intensive lifestyle intervention: the look AHEAD study", Obesity 22(1), 2014, pp. 5 to 13, doi 10.1002/oby.20662. Read: the abstract. Supports: the eight-year proportions and "clinically meaningful".
11. R. R. Wing and S. Phelan, "Long-term weight loss maintenance", American Journal of Clinical Nutrition 82(1 Suppl), 2005, pp. 222S to 225S, doi 10.1093/ajcn/82.1.222S. Read: the abstract. Supports: the 33 kg and five years, the reported habits, the 20 percent figure and its definition, and the conclusion. That the registry can't show how common maintenance is or what causes it is this course's reading.
12. J. P. H. Wilding and colleagues, "Once-Weekly Semaglutide in Adults with Overweight or Obesity", New England Journal of Medicine 384(11), 2021, pp. 989 to 1002, doi 10.1056/NEJMoa2032183. Read: the abstract. Supports: the design, the weight results and the funder. That the drugs act on appetite, and the comparison with the diet trials, are this course's.
13. J. P. H. Wilding and colleagues, "Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension", Diabetes, Obesity and Metabolism 24(8), 2022, pp. 1553 to 1564, doi 10.1111/dom.14725. Read: the abstract. Supports: the 327 participants, that the lifestyle programme stopped too, the exploratory status and the two-thirds regain.
14. World Health Organization, "Obesity and overweight" fact sheet, who.int, dated 8 December 2025. Read: the page in full, 24 September 2026. Supports: the three quotations and the December 2025 GLP-1 guideline, which this course hasn't read. That the responsibility sentence is a value position is this course's reading.
15. F. Rubino and colleagues, "Definition and diagnostic criteria of clinical obesity", Lancet Diabetes & Endocrinology 13(3), 2025, pp. 221 to 262, doi 10.1016/S2213-8587(24)00316-4. Read: the executive summary. Supports: the preclinical obesity definition. That WHO's "disease" is a classification this commission draws differently is this course's reading.