Patterns, and the guidelines that bundle them

On 13 June 2018 the New England Journal of Medicine retracted the best-known diet trial of the decade and, the same day, published it again.³ PREDIMED had assigned 7,447 people in Spain to a Mediterranean diet with extra olive oil, the same diet with extra nuts, or advice to cut fat. Part of it turned out not to have been randomised. Rerun, the answer barely moved; what moved was the kind of study it was. Most advice is about whole ways of eating, not single nutrients, and national guides bundle those patterns into advice. The 2025-2030 US guidelines are the most argued-over document in this course. Here's what each pattern's best evidence shows, where the guides agree, and what the 2026 dispute is about, in words each side would sign.

Guidelines and the arguments about them change, so every claim here about the 2026 US guidelines is marked "as of September 2026".

Testing a whole diet

Lesson 1 said you cannot change one thing. A trial of a single nutrient fights that; a trial of a pattern gives in, assigning a whole way of eating at once. That buys a question closer to how people eat, and it costs explanation: if the package works, the trial can't say which part did it. That's this course's reasoning from lesson 1's point, and the original DASH trial shows both halves.

Appel and colleagues, in 1997, fed 459 adults with systolic pressure under 160 and diastolic 80 to 95 mm Hg a control diet "low in fruits, vegetables, and dairy products, with a fat content typical of the average diet in the United States" for three weeks, then randomised them for eight weeks to the control diet, a diet rich in fruit and vegetables, or a "combination" diet that added low-fat dairy and cut saturated and total fat.⁹ All food was provided, and "Sodium intake and body weight were maintained at constant levels." This course read the abstract.

Holding sodium and weight steady means whatever lowered blood pressure was in the food, and the middle arm shows fruit and vegetables alone got part of the way. But the trial can't say whether the dairy, the lower fat, the fibre or the potassium did the rest, and eight weeks of blood pressure isn't a heart attack. So a pattern trial is good evidence for advice ("eat like this") and weak evidence for mechanism ("this nutrient is why"). Lesson 7 read the later trial that crossed DASH with sodium levels.

The Mediterranean trials

PREDIMED, before and after

The 2013 paper by Estruch and colleagues reported 7,447 people aged 55 to 80, 57 percent women, at high cardiovascular risk but without heart disease, at 11 sites.¹ They got "quarterly individual and group educational sessions and, depending on group assignment, free provision of extra-virgin olive oil, mixed nuts, or small nonfood gifts". The control arm got "advice to reduce dietary fat". The trial stopped early after a median of 4.8 years, with 288 major cardiovascular events (heart attacks, strokes and cardiovascular deaths). Hazard ratios, lesson 2's measure of relative risk over time, were 0.70 (0.54 to 0.92) for olive oil and 0.72 (0.54 to 0.96) for nuts. The conclusion: "Among persons at high cardiovascular risk, a Mediterranean diet supplemented with extra-virgin olive oil or nuts reduced the incidence of major cardiovascular events."¹

In 2018 the same authors wrote: "We subsequently identified protocol deviations, including enrollment of household members without randomization, assignment to a study group without randomization of some participants at 1 of 11 study sites, and apparent inconsistent use of randomization tables at another site."² Harvard's Nutrition Source, read in full, adds that at the one site "several clinics instead of individual patients" were assigned.³ So they withdrew the paper "and now report revised effect estimates based on analyses that do not rely exclusively on the assumption that all the participants were randomly assigned."² Both abstracts were read for this course.

Secondary accounts link the discovery to a 2017 paper by John Carlisle that tested whether the groups' starting characteristics in 5,087 trials looked like chance; its abstract doesn't name PREDIMED, and this course didn't verify the link.⁴ This course found no allegation of fraud against PREDIMED.

The 2018 estimates, "adjusting for baseline characteristics and propensity scores", were 0.69 (0.53 to 0.91) and 0.72 (0.54 to 0.95), and "Results were similar after the omission of 1588 participants whose study-group assignments were known or suspected to have departed from the protocol."² (A propensity score is each person's estimated chance of being in a group, given what was measured about them, used to balance the groups.) The 1,588 are about 21 percent of 7,447, by this course's arithmetic.

What a reader should now say about it

Step 1: the number survived. Every interval on the chart is still below 1, and the estimates moved by a hundredth or not at all.

Step 2: the kind of evidence changed. For up to about a fifth of the people, those whose assignment was known or suspected to have broken protocol, chance may not have decided who ate what, so for them the result rests on adjustment, which can only balance what was measured. John Ioannidis, in a 2018 viewpoint read in full, granted that the reanalysis "showed results similar to those of the initially reported findings; however, the study should no longer be considered a randomized trial", and added that "the trial showed no survival benefit".⁵ So the two readings agree on the number. They differ on what kind of evidence it now is, and on how much the sentence about the 1,588 repairs that. Notice the verbs, too. In 2013 the diet "reduced the incidence of major cardiovascular events"; in 2018 the incidence "was lower among those assigned to a Mediterranean diet supplemented with extra-virgin olive oil or nuts than among those assigned to a reduced-fat diet."¹² The first claims a cause, the second describes a difference. That reading is this course's.

Step 3: the absolute size. The 2018 abstract gives crude event rates of 3.8 percent with olive oil, 3.4 percent with nuts and 4.4 percent in the control arm, over a median of 4.8 years.² Work out the differences before opening the box.

Step 4: what else was graded. The 2019 Cochrane review by Rees and colleagues, read at abstract level, graded PREDIMED outcome by outcome.⁶ For cardiovascular deaths it found low-quality evidence of "little or no effect" (hazard ratio 0.81, 0.50 to 1.32), and the same for deaths from any cause (1.0, 0.81 to 1.24). For strokes it found a reduction at moderate quality, 0.60 (0.45 to 0.80), "a decrease from 24/1000 to 14/1000". Across all the trials it reviewed, "there is still some uncertainty regarding the effects of a Mediterranean-style diet on clinical endpoints".

So a reader should now say something like: PREDIMED found fewer major cardiovascular events, mostly strokes, in people at high risk given olive oil or nuts, 0.6 to 1 fewer per 100 over about five years against low-fat advice; the result survived a reanalysis that no longer rests wholly on randomisation, and it didn't show fewer deaths.

Lyon and CORDIOPREV

Two trials tested the pattern in people who already had heart disease, which cardiologists call secondary prevention. The Lyon Diet Heart Study randomised patients after a first heart attack to a Mediterranean-type or a "prudent Western-type diet" for a mean of 46 months; Cochrane counts 605 patients and describes the intervention as advice "and supplemental canola margarine".⁶⁷ Cardiac death and non-fatal heart attack occurred 14 times against 44 (the abstract doesn't give the group sizes), and an interim analysis had already shown "a striking protective effect".⁷ Cochrane graded its reductions in cardiovascular and total deaths (adjusted hazard ratios 0.35 and 0.44) as low-quality evidence.⁶

CORDIOPREV, published by Delgado-Lista and colleagues in 2022, randomised 1,002 patients with coronary heart disease at one hospital in Córdoba to a Mediterranean or a low-fat diet, both taught by dietitians, for seven years.⁸ The combined outcome (heart attacks, procedures to reopen arteries, ischaemic strokes, peripheral artery disease and cardiovascular deaths) occurred in 87 of 502 people against 111 of 500, crude rates of 28.1 against 37.7 per 1,000 person-years, with adjusted hazard ratios from 0.719 (0.541 to 0.957) to 0.753 (0.568 to 0.998). In men, 16.2 against 22.8 percent had an event; there were 175 women, "for whom no difference was found between groups." The authors: "In secondary prevention, the Mediterranean diet was superior to the low-fat diet in preventing major cardiovascular events." This course read both abstracts. Read the upper ends of those intervals against 1, as lesson 6 did: 0.957 and 0.998. On its own, CORDIOPREV only just clears the line. The case for combined events is stronger because three trials, each counting a slightly different bundle of events, point the same way. For deaths the evidence is thinnest: little or no effect in PREDIMED, and only low-quality evidence from Lyon.

Vegetarian and vegan patterns

The Academy of Nutrition and Dietetics, an American association of dietitians, puts its position in one paragraph: "It is the position of the Academy of Nutrition and Dietetics that appropriately planned vegetarian, including vegan, diets are healthful, nutritionally adequate, and may provide health benefits for the prevention and treatment of certain diseases."¹⁰ It says they suit every stage of life, and adds: "Vegans need reliable sources of vitamin B-12, such as fortified foods or supplements." Its environmental claim, and the ethical reasons for eating this way, are beyond this course.

The weight sits on "appropriately planned". The 2025-2030 US guidelines, as of September 2026, put the caution: "Vegetarian diets often fall short in vitamins D and E, choline, and iron, whereas vegan diets show broader shortfalls in vitamins A, D, E, B6, and B12; riboflavin; niacin; choline; calcium; iron; magnesium; phosphorus; potassium; zinc; and protein."[21] In the next column the same page advises: "prioritize targeted supplementation, diversify plant protein sources for amino acid balance", which is planning advice. On this course's reading both can be true, one about planned diets and one about diets as eaten, though they leave different impressions.

On B12 every source agrees. The US Office of Dietary Supplements says natural sources "are limited to animal foods" and that fortified foods and supplements "can substantially reduce the risk of deficiency"; for babies of vegan mothers it warns: "Undetected and untreated vitamin B12 deficiency in infants can result in neurological damage, failure to thrive, developmental delays, and anemia".¹² For pregnancy and children this is the callout's territory: see a doctor or a registered dietitian. Lesson 10 returns to B12.

The best long cohort on outcomes is EPIC-Oxford, reported by Tong and colleagues in 2019: 48,188 people in the UK, with a large share of non-meat eaters, followed for 18 years.¹¹ This course read the abstract.

The heart advantage shrank after adjustment for self-reported risk factors such as blood pressure and diabetes, to "hazard ratio 0.90, 95% confidence interval 0.81 to 1.00 in vegetarians with all adjustments", which suggests part of it runs through those risk factors. The stroke finding didn't: "The associations for stroke did not attenuate after further adjustment of disease risk factors."¹¹ It's a cohort, so both carry lesson 1's caveat, and this course didn't read the paper's discussion of the stroke result, so it offers no mechanism.

Low-carbohydrate patterns

Lesson 4 promised the low-carbohydrate evidence in diabetes. First, the American Diabetes Association, which has accepted the pattern since at least 2019. Its consensus report that year, by Evert and colleagues, read in its eating-pattern and carbohydrate sections, says: "For select adults with type 2 diabetes not meeting glycemic targets or where reducing antiglycemic medications is a priority, reducing overall carbohydrate intake with low- or very low-carbohydrate eating plans is a viable approach."¹³ Its 2026 standards list "Mediterranean, DASH, low-fat, carbohydrate-restricted, vegetarian, and vegan eating patterns" together.[14]

The same report carries the safety line behind this course's callout, for very low carbohydrate plans: "consultation with a knowledgeable practitioner at the onset is necessary to prevent dehydration and reduce insulin and hypoglycemic medications to prevent hypoglycemia."¹³ Hypoglycaemia is blood sugar falling too low. It also flags "theoretical concerns regarding use of VLC eating plans in people with chronic kidney disease, disordered eating patterns, and women who are pregnant", where "further research is needed". (VLC is very low carbohydrate.) Those are three of the groups this course's callout names.

The proponents' own trial

The Virta Health trial, reported by Hallberg and colleagues in 2018 with Stephen Phinney and Jeff Volek among the authors, put 262 adults with type 2 diabetes into remote coaching, doctor-managed medicines and a diet aimed at nutritional ketosis (so little carbohydrate that the body runs largely on fat), against 87 in usual care.[15] At a year, average HbA1c, a measure of blood sugar over about three months, fell from 7.6 to 6.3 percent in the programme group, and "Insulin therapy was reduced or eliminated in 94% of users", meaning people who had been taking insulin. LDL cholesterol rose 10 percent, while "apolipoprotein B was unchanged" (another lipid measure; this course has no source on how to weigh the two, so it reports both). At two years the authors reported "resolution of diabetes (reversal, 53.5%; remission, 17.6%) in the CCI group", CCI being the programme.[16] The abstract doesn't say how its reversal and remission differ, so treat them as the authors' own terms. This course read both abstracts.

Now the design, which the paper states plainly: "an open-label, non-randomized, controlled, before-and-after 1-year study". Open-label means everyone knew which group they were in, as in lesson 7's salt trial. The usual-care group wasn't assigned by chance; as this course's research summarises it, it was drawn from people who hadn't opted in. The programme bundles coaching, close medical care and the diet, so the diet's share can't be separated: the pattern trial's limit in its sharpest form. And the abstract ends: "Funding Virta Health Corp."[15]

The randomised evidence

Goldenberg and colleagues pooled 23 randomised trials in 2021, read at abstract level.[17] At six months, remission (HbA1c under 6.5 percent) was commoner on low-carbohydrate diets, "76/133 (57%) v 41/131 (31%)", but "smaller, non-significant effect sizes occurred when a remission definition of HbA 1c <6.5% without medication was used", and "At 12 months, data on remission were sparse, ranging from a small effect to a trivial increased risk of diabetes." An HbA1c under 6.5 reached while still taking medicine is partly the medicine's doing, so on this course's reading the drug-free definition is the harder test of the diet, and on it the advantage shrank. Their conclusion: "On the basis of moderate to low certainty evidence, patients adhering to an LCD for six months may experience remission of diabetes without adverse consequences." (LCD is low-carbohydrate diet.) The ADA's 2026 standards agree: carbohydrate-restricted patterns "were effective in reducing A1C in the short term (<6 months), with less difference in eating patterns beyond 1 year".[14]

In Keto-Med (2022), 33 people with prediabetes or type 2 diabetes ate a ketogenic and a Mediterranean-style diet for 12 weeks each, in random order; "The HbA1c values did not differ between diets at 12 weeks", LDL rose about 10 percent on the ketogenic diet, and "Twelve-week follow-up data suggest the Med-Plus is more sustainable."[18] Its lead author, C. D. Gardner, co-wrote the ADA report, so the two aren't independent voices.

At the far end, the carnivore diet's one study here is a 2021 survey of 2,029 people eating only animal foods, by Lennerz and colleagues, with the low-carbohydrate researcher David Ludwig among the authors.[19] The course project's worked model traces it step by step, including what a survey of people still on a diet can't see.

Five guides side by side

This course read five guides. Four are in the table; you'll add Canada's in the exercise. "Not given" means the passages this course read don't give a figure, not that the guide is silent.

Sources: [20], [21], [29], [31]. The UK column combines the NHS Eatwell page and Public Health England's 2016 table, whose saturated fat and sugar figures are calculated at "11% food energy" and "5% food energy".[29]

Where they agree. Vegetables and fruit, whole grains in place of refined, less added sugar, less salt: every column points the same way, even where the numbers differ. Three cap saturated fat at 10 percent of energy and the UK sits at 11. Two of the four columns name processed food, and so does Canada, which you'll add. The 2026 US launch page says "For the first time, official U.S. guidance calls on Americans to avoid highly processed food."[21] True of US guidance; Canada's 2019 guide had already said "Limit highly processed foods".[30]

Where they differ. Protein amount, where the 2025-2030 US text stands alone at 1.2 to 1.6 g per kg; protein source, where the UK and WHO lean towards plants and the new US text lists animal sources first; dairy fat, where only the new US text recommends full-fat and WHO names reduced-fat. Alcohol is the odd row: the old US edition is the only column with a number, and the new one dropped it. On the other disputed rows the outlier is one document, the newest.

WHO's column is its fact sheet as revised on 26 January 2026, which now grants that "low and very low carbohydrate diets are used to treat certain conditions" while putting most people's carbohydrate at "approximately 45–75% of total daily energy".[31]

The UK's salt figure carries a footnote: "These target salt intakes do not represent ideal or optimum consumption levels, but achievable population goals."[29] A guideline number is often a target for a population, lesson 5's point about the RDA.

The 2025-2030 US Dietary Guidelines, as of September 2026

Lessons 5 to 8 quoted lines from these guidelines. Here is the whole, and the argument over it.

What the document is

The Dietary Guidelines for Americans, 2025-2030 were released on 7 January 2026, signed by the health and agriculture secretaries, Robert F. Kennedy, Jr. and Brooke L. Rollins.[21] The document is ten pages, against the previous edition's 164, with a 90-page companion, The Scientific Foundation for the Dietary Guidelines for Americans, 2025–2030. This course read the first whole and parts of the second.[21][22] It opens: "These Guidelines mark the most significant reset of federal nutrition policy in our nation's history. The message is simple: eat real food."[21] A graphic its authors call the New Pyramid replaced MyPlate, the plate-shaped graphic of the 2020-2025 edition; this course read descriptions of it, not the image.

Beyond the lines earlier lessons quoted (protein, fats, sugars, sodium, processed food), three more changes matter here. Dairy: "When consuming dairy, include full-fat dairy with no added sugars", three servings a day, where the old text said fat-free or low-fat. Alcohol: "Consume less alcohol for better overall health.", with no number. And: "Individuals with certain chronic diseases may experience improved health outcomes when following a lower carbohydrate diet."[21] The American Nutrition Association, on 9 January, welcomed "the role of lowered carbohydrates in reversing some chronic diseases" as part of "a meaningful advancement".[26]

What supporters and critics agree on

Start here, because the headlines didn't. The American Heart Association, on the day of release: "The American Heart Association welcomes the new dietary guidelines and commends the inclusion of several important science-based recommendations, notably the emphasis on increasing intake of vegetables, fruits and whole grains while limiting consumption of added sugars, refined grains, highly processed foods, saturated fats and sugary drinks."[23] Frank Hu, who chairs Harvard's nutrition department, quoted by Harvard's Nutrition Source on 9 January: "I think the new Guidelines move in the right direction by reinforcing the importance of reducing added sugars and cutting back on refined grains and other highly processed foods ... However, there appear to be several contradictions within the DGAs and between the DGAs and the new pyramid."[27]

In this course's summary: less added sugar, fewer refined grains and highly processed foods, more vegetables, fruit and whole grains, and the saturated fat cap kept at 10 percent. Those are the rows where the table agrees.

Where they part

Protein. For: the guidelines' heading is "Prioritize Protein Foods at Every Meal", and realfood.gov says: "We are ending the war on protein. Every meal must prioritize high-quality, nutrient-dense protein from both animal and plant sources, paired with healthy fats from whole foods such as eggs, seafood, meats, full-fat dairy, nuts, seeds, olives, and avocados."[21] The Scientific Foundation objected that the advisory committee "consistently advocated plant-based dietary patterns, deprioritized animal-sourced proteins, and favored high linoleic acid vegetable oils", and called its plan to list beans, peas and lentils ahead of meat "a symbolic reordering lacking scientific justification".[22] Against: Hu warns that "Substantially raising overall protein intake without distinguishing between different protein sources may have unintended long-term health implications", and the AHA urges people "to prioritize plant-based proteins, seafood and lean meats".[23][27]

This course didn't read the Scientific Foundation's protein chapter. A critical 2026 commentary by Pasiakos and colleagues in the Journal of Nutrition, read at abstract level, describes what the figure rests on: "a rapid systematic review of randomized controlled trials focused on weight-management outcomes and, to a lesser extent, a narrative review on nutrient adequacy". It calls that basis "insufficiently comprehensive", and grants: "Recognizing protein as a component of a healthy diet is consistent with recent editions and a strength of the 2025-2030 DGA."[37] Lesson 5's evidence cuts both ways. The PROT-AGE expert group recommends at least 1.0 to 1.2 g per kg for people over 65, and in trials of people lifting weights, extra protein added no further lean mass beyond about 1.6, the top of the new range. But KDIGO advises against more than 1.3 g per kg in chronic kidney disease at risk of progression.

Full-fat dairy, butter and tallow. For: the Scientific Foundation argues that "emphasis on restricting saturated fat in dietary guidelines and authorized health claims may have inadvertently encouraged selection of processed products, such as low- and nonfat yogurts with added sugars, artificial sweeteners, emulsifiers, thickeners, and other chemical additives."[22] Against: an expert analysis for the American College of Cardiology, published on 27 January and read in full, says "evidence from feeding trials and/or cohort studies does not support promotion of butter or beef tallow. Moreover, if the newly recommended servings of animal protein foods (e.g., meats and whole-fat dairy) and proposed 'healthy fats' (e.g., butter and beef tallow) are incorporated into the diet, saturated fat intake will exceed 10% of daily calories."[24] Harvard's Nutrition Source worked a 2,000-calorie day: three full-fat dairy servings come to about 17 grams of saturated fat against a ceiling of about 22, before any butter.[27] The same ACC analysis grants that "dairy food consumption has not been associated with increased CV risk".[24]

Red meat and the picture. The AHA is "concerned that recommendations regarding salt seasoning and red meat consumption could inadvertently lead consumers to exceed recommended limits for sodium and saturated fats, which are primary drivers of cardiovascular disease."[23] The American Nutrition Association, which had welcomed the low-carbohydrate line, objected to "the placement of red meat and dairy products at the 'top of the chart', appearing above foods representing a Mediterranean diet, which has the largest body of evidence in preventing and treating chronic disease."[26] For realfood.gov the pyramid "prioritizes nutrient-dense foods and reduces reliance on highly processed products".[21]

Alcohol. The old limit of two drinks a day for men and one for women is gone. Criticism and welcomes reached this course only as unread search results, so it quotes neither side. "Consume less alcohol" still says less is better, and the text lists people who "should completely avoid alcohol", pregnant women among them.[21] Lesson 10 has the evidence.

The process dispute

The sharpest disagreement is about how the guidelines were made. The 2025 Dietary Guidelines Advisory Committee, a panel of scientists, wrote a scientific report that preceded them. The Scientific Foundation says the administration "implemented an independent evidence review process to address and correct deficiencies identified in the Scientific Report of the 2025 Dietary Guidelines Advisory Committee (DGAC Report), which framed its analysis through a health equity lens", and that "the Trump Administration determined that adopting the DGAC Report would not meet the American public's need for objective, evidence-based nutrition guidance."[22] Its reviewers "were selected through a federal contracting process", and NIH's Office of Nutrition Research "coordinated an external peer review process, assigning two independent reviewers to each scientific review". Its data were compiled by Christopher Ramsden (lesson 6). It names its nine review authors (Ty Beal, Benjamin Bikman, J. Thomas Brenna, Michael Goran, Heather Leidy, Ameer Taha, Jeff Volek, Daisy Zamora and Donald Layman) and prints their disclosures.[22]

Against the process. The American Society for Nutrition, which supported the broad eating pattern, said on 9 January: "departing from the established scientific review process undermines confidence in the DGAs and nutrition science", and "There is a lack of transparency regarding the methods, approaches, and objectives of the newly introduced scientific reviews and the timeframe in which they were completed."[25] Deirdre Tobias of Harvard, a member of the 2025 committee, in a Harvard Chan News interview quoted by the Nutrition Source on 9 January: "As of today, there has not been transparency in who wrote the new DGAs."[27]

For it. The administration's case is the paragraph above: the committee's report was deficient, and the replacement was peer reviewed and names its authors. Nina Teicholz, a science journalist with a doctorate who founded the Nutrition Coalition, campaigned against the old process for years. In a September 2022 post about the guidelines then in force, she wrote that they "cannot be guaranteed to reflect trustworthy advice for the urgent task of combating obesity, diabetes or any other chronic disease", and cited a 2017 National Academies call to "improve transparency" and "strengthen scientific rigor".[28] Zoë Harcombe, a researcher who favours lower carbohydrate eating, wrote in January 2026 that the new guidelines "bore almost no resemblance to the Scientific Report which preceded them", and called it a "monumental change".[28]

Harcombe and the ASN agree that the guidelines departed from the committee's report, and part over whether that was a repair or a breach. The ASN also said it "supports the respected ASN members who served as scientific review authors".[25]

As of September 2026 the process question is open, and this course doesn't answer it. Whether an administration should set aside its advisory committee's report is partly about evidence and mostly about who should decide, by what procedure, with what accountability, which are value questions informed people answer differently. Journal commentaries followed through 2026. This course read three, all critical (two on fats, and Pasiakos on protein), and none from the guidelines' side.

In this course's view, what would settle the empirical parts is long trials counting heart attacks and deaths that compare an animal-protein-led pattern with a plant-protein-led one at the same saturated fat cap. There's none; PREDIMED and CORDIOPREV are the nearest, and neither tested that contrast.

Blue Zones: a claim that travelled

In 2004 Poulain and colleagues mapped a mountainous part of Sardinia with unusually many validated centenarians, especially men, and called it the "Blue Zone".[32] The abstract doesn't mention diet; it floats inbreeding in isolated villages as "An alternative and interesting hypothesis". By 2016, in a paper by Dan Buettner and Sam Skemp, an expedition had become "the discovery of the 5 places around the world where people consistently live over 100 years old", distilled into "the Power 9" and carried into Blue Zones Project communities that "have been able to increase life expectancy", a commercial programme's claim with no data in the abstract.[33]

Saul Newman's 2019 preprint (not peer reviewed; abstract read) argues for "a primary role of fraud and error in generating remarkable human age records", and finds that "the designated 'blue zones' of Sardinia, Okinawa, and Ikaria corresponded to regions with low incomes, low literacy, high crime rate and short life expectancy relative to their national average."[34] It won an Ig Nobel prize in September 2024.[35]

Six researchers answered in an October 2024 "Demographers' Statement", read in full: "In Sardinia, for example, we double-checked the age of every centenarian in the Blue Zone villages using: Civil status databases dating back to 1866". They call Newman's use of Sardinia-wide statistics "misleading", since the zone is a small rural area, and concede: "we have publicly acknowledged that Okinawa's longevity rates have declined with the arrival of modern diets and transportation."[36] The statement is hosted on bluezones.com, the Blue Zones company's site. As of September 2026 the critique was still a preprint, going by a July 2026 news report that this course read only as a summary.

Trace the claim as Sleep lesson 7 traced a bestseller's, and it splits into two questions that travelled as one. On this course's reading of both sides: are the ages real? Contested, least for the Sardinian villages, where the validation described is the most careful. Did the diet cause them? That's never been tested. The Power 9 describe what long-lived people did, gathered afterwards, with no comparison group: Logic and Argument lesson 5's method of difference with the difference missing. The second question fails whatever the answer to the first.

Who paid

None of what follows is a reason to dismiss a finding. CORDIOPREV was funded partly by two olive oil foundations; PREDIMED by Spain's Instituto de Salud Carlos III "and others".²⁸ Virta Health funded the trial of its own programme.[15] The Scientific Foundation's disclosures include Brenna's "National Cattlemen's Beef Association/ Texas Beef Council. Consulting and research grant."[22], and lesson 6 listed the matching interests on the mainstream side of the fat dispute. The Blue Zones reply is hosted by the company that sells the programme.[36] Brenna's line is here because the Foundation printed its authors' disclosures. This course didn't read the other papers' disclosures or those of the AHA, ACC, ASN or ANA, so a missing line is not a finding.

What people get wrong

"The Mediterranean diet was proven in a trial that was later retracted, so it's debunked." The estimate survived almost unchanged, and CORDIOPREV points the same way.²⁸ What changed is that PREDIMED now rests partly on adjustment.

"The Mediterranean diet is proven to save lives." The trials found fewer combined events, mostly strokes in PREDIMED. Cochrane found little or no effect on deaths in PREDIMED, and graded Lyon's reduction in deaths low-quality.⁶

"Vegetarian diets are healthier in every outcome." EPIC-Oxford found less ischaemic heart disease and more haemorrhagic stroke.¹¹

"A vegetarian or vegan diet can't be nutritionally adequate." The Academy says appropriately planned ones are, at every stage of life. On this course's reading the US guidelines' shortfalls describe diets as eaten, and the same page advises targeted supplementation. B12 needs a fortified food or supplement.¹⁰¹²[21]

"Low-carbohydrate eating is a fad the experts reject." The ADA has accepted low and very low carbohydrate plans for some adults with type 2 diabetes since at least 2019.¹³[14]

"Low-carbohydrate eating is proven best for diabetes." The same body reports the advantage fading beyond about a year, and its caution about medicines is a safety line.¹³[14]

"The guides all disagree." On vegetables, fruit, whole grains, sugar and salt they agree.

"The 2026 US guidelines were the first to tell anyone to avoid highly processed food." The first US edition to; Canada said "Limit highly processed foods" in 2019.[30]

"The 2026 US guidelines scrapped the limit on saturated fat." They kept it: "In general, saturated fat consumption should not exceed 10% of total daily calories." Harvard's Nutrition Source warned that the new pyramid might leave readers thinking the limit had gone. The dispute is over whether the butter, tallow and full-fat dairy advice fits under the cap.[21][24][27]

"Blue Zones prove a diet makes people live to 100." The ages are disputed, and diet as the cause was never tested.[32][34][36]

Practice

Connections

Back. Lesson 1's "you cannot change one thing" is why pattern trials exist. Lesson 4's promised low-carbohydrate evidence is Virta, design stated, with the randomised trials beside it. Lessons 5 to 8 are the parts these patterns and guides bundle.

Forward. Lesson 10 takes supplements, B12 for vegans among them, and alcohol.

Go deeper

• Dietary Guidelines for Americans, 2025-2030, free. Read here in full. Ten pages, worth reading before anyone's opinion of them.
• Canada's Food Guide, free. Read here through archived copies of two pages. The guide you rebuilt in the exercise.
• Rees and colleagues, 2019, the Cochrane review of Mediterranean-style diets. Read here: the abstract. PREDIMED and Lyon graded outcome by outcome.
• Tong and colleagues, 2019, EPIC-Oxford, free at the BMJ. Read here: the abstract. A benefit and a trade-off, side by side.

Sources

1. R. Estruch and colleagues (PREDIMED), New England Journal of Medicine 368(14), 2013, pp. 1279 to 1290, doi 10.1056/NEJMoa1200303. Retracted. Read: the abstract.
2. R. Estruch and colleagues (PREDIMED, republished), New England Journal of Medicine 378(25), 2018, e34, doi 10.1056/NEJMoa1800389. Read: the abstract. The retraction notice was not read.
3. Harvard T.H. Chan School of Public Health, The Nutrition Source, explainer on the PREDIMED retraction, 22 June 2018. Read: in full.
4. J. B. Carlisle, on non-random sampling in 5,087 trials, Anaesthesia 72(8), 2017, doi 10.1111/anae.13938. Read: the abstract, which does not name PREDIMED.
5. J. P. A. Ioannidis, viewpoint, JAMA 320(10), 2018, doi 10.1001/jama.2018.11025. Read: the full text (2 pages).
6. K. Rees and colleagues, Cochrane review of Mediterranean-style diets, 2019, CD009825. Read: the abstract. Its "7747" for PREDIMED differs from the trial papers' 7,447; the lesson uses 7,447.
7. M. de Lorgeril and colleagues, final report of the Lyon Diet Heart Study, Circulation 99(6), 1999, doi 10.1161/01.CIR.99.6.779. Read: the abstract. The 605 is from [6].
8. J. Delgado-Lista and colleagues (CORDIOPREV), Lancet 399(10338), 2022, doi 10.1016/S0140-6736(22)00122-2. Read: the abstract, with its funding line.
9. L. J. Appel and colleagues (DASH), New England Journal of Medicine 336(16), 1997, doi 10.1056/NEJM199704173361601. Read: the abstract.
10. V. Melina, W. Craig and S. Levin, Academy of Nutrition and Dietetics position on vegetarian diets, Journal of the Academy of Nutrition and Dietetics 116(12), 2016, doi 10.1016/j.jand.2016.09.025. Read: the abstract, which is the position statement.
11. T. Y. N. Tong and colleagues (EPIC-Oxford), BMJ 366, 2019, l4897, doi 10.1136/bmj.l4897. Read: the abstract.
12. NIH Office of Dietary Supplements, "Vitamin B12: Health Professional Fact Sheet", updated 2 July
    2025. Read: through an Internet Archive copy.
13. A. B. Evert and colleagues, ADA consensus report on nutrition therapy, Diabetes Care 42(5), 2019, doi 10.2337/dci19-0014. Read: the full-text sections on eating patterns, macronutrients and carbohydrate.
14. American Diabetes Association, Standards of Care in Diabetes, 2026, section 5, Diabetes Care, doi 10.2337/dc26-S005. Read: the carbohydrate and eating-pattern sections.
15. S. J. Hallberg and colleagues (Virta, one year), Diabetes Therapy 9(2), 2018, doi 10.1007/s13300-018-0373-9. Read: the abstract.
16. S. J. Athinarayanan and colleagues (Virta, two years), Frontiers in Endocrinology 10, 2019, doi 10.3389/fendo.2019.00348. Read: the abstract.
17. J. Z. Goldenberg and colleagues, low and very low carbohydrate diets for type 2 diabetes remission, BMJ 372, 2021, m4743, doi 10.1136/bmj.m4743. Read: the abstract.
18. C. D. Gardner and colleagues (Keto-Med), American Journal of Clinical Nutrition 116(3), 2022, doi 10.1093/ajcn/nqac154. Read: the abstract.
19. B. S. Lennerz and colleagues, survey of 2,029 adults on a carnivore diet, Current Developments in Nutrition 5(12), 2021, doi 10.1093/cdn/nzab133. Read: the abstract.
20. USDA and HHS, Dietary Guidelines for Americans, 2020-2025, December 2020, replaced by [21]. Read: the Executive Summary and the fibre section, from an Internet Archive copy.
21. HHS and USDA, Dietary Guidelines for Americans, 2025–2030, January 2026, PDF, and realfood.gov. Read: the whole document and the launch page. The release date is from [23].
22. HHS and USDA, The Scientific Foundation for the Dietary Guidelines for Americans, 2025–2030, January 2026, PDF. Read: the front matter, the author list, the disclosures, and Chapter 5, "Fats and Oils". Its protein chapter and appendices were not read.
23. American Heart Association, statement on the new guidelines, 7 January 2026. Read: in full.
24. M. Aggarwal, K. E. Aspry, P. M. Kris-Etherton, A. M. Freeman, K. A. Williams Sr and E. Gianos, expert analysis for the American College of Cardiology, acc.org, 27 January 2026. Read: in full.
25. American Society for Nutrition, statement, 9 January 2026. Read: in full.
26. American Nutrition Association, statement, 9 January 2026. Read: in full.
27. Harvard T.H. Chan School of Public Health, The Nutrition Source, on the 2025-2030 guidelines, 9 January 2026. Read: in full. Tobias's words were read here, not in the interview they come from.
28. Nina Teicholz, "Dietary Guidelines are Unscientific, Outdated", Nutrition Coalition, 8 September 2022, and Zoë Harcombe's post of January 2026. Read: both pages. The 2017 National Academies report was not read.
29. NHS, "The Eatwell Guide" (modified 9 October 2025), and Public Health England, Government Dietary Recommendations, 2016, via the GOV.UK Eatwell page. Read: the NHS page in full, and tables 1 and 2 with their footnotes.
30. Health Canada, Canada's Food Guide, 2019. Read: Internet Archive 2025 copies of two pages, in full.
31. World Health Organization, "Healthy diet", fact sheet, revised 26 January 2026, who.int. Read: in full.
32. M. Poulain and colleagues, the AKEA study, Experimental Gerontology 39(9), 2004, doi 10.1016/j.exger.2004.06.016. Read: the abstract.
33. D. Buettner and S. Skemp, American Journal of Lifestyle Medicine 10(5), 2016, doi 10.1177/1559827616637066. Read: the abstract.
34. S. J. Newman, on supercentenarian and remarkable age records, bioRxiv preprint, 2019, doi 10.1101/704080. Read: the abstract of the version Europe PMC indexes. Not peer reviewed.
35. University College London, press release on the 2024 Ig Nobel prize in demography. Read: through a mirror.
36. G. Pes, L. Rosero-Bixby, C. Chrysohoou, C. Stefanadis, B. Willcox and C. Willcox, "Demographers' Statement", October 2024, bluezones.com. Read: in full. The July 2026 news report on the preprint's status was read only as a summary.
37. S. M. Pasiakos and colleagues, "Perspectives on the Protein Recommendations in the 2025-2030 Dietary Guidelines for Americans", Journal of Nutrition, 2026, doi 10.1016/j.tjnut.2026.101754. Read: the abstract.