Full mechanism, in text

Same calories, yet people eat more

1 · design

The trial housed 20 adults in a metabolic ward for 2 weeks each of ultra-processed and unprocessed diets, matched for nutrients, eating as much as they wanted.Hall et al 2019 Cell Metabolism — the first randomized trial to put ultra-processed food into a metabolic ward:

Institution: NIH NIDDK metabolic ward (metabolic ward, not self-reported food questionnaire)

N = 20 healthy adults (10 men + 10 women, 27)

Design — strict crossover:

· Each person admitted for 4 weeks, diet switched midway
· One 2 weeks eating fully ultra-processed (), one 2 weeks fully unprocessed (matched whole foods)
· The two periods ran back to back, in random order

Key controls — why this trial matters:

· The two diets were precisely matched on:
Total calories (provided)Protein (g) ✓Carb (g) ✓Fat (g) ✓Sugar (g) ✓Sodium (mg) ✓Fibre (g) ✓· Main difference = ultra-processed or not

Intake rules:

· Subjects ate ad libitum (eat as much as desired, no calorie cap)
· Three meals + snacks per pre-set menu, but quantity was free

Measurements:

· Body weight (daily)
· Total intake (weighed leftovers)
· Hunger / satiety hormones (PYY, ghrelin, , leptin)
· Metabolic rate (doubly labelled water)
· Appetite scales

Why this design matters: earlier UPF studies were essentially all observational — people who eat more UPF are heavier, but cause and effect could not be separated. This trial matched the nutrients and looked at the degree of processing itself.

2 · result

In the weeks people ate about 508 kcal a day more and gained 0.9 kg, while in the unprocessed weeks they lost 0.9 kg.Intake difference:

· On the diet people ate 508 ± 106 kcal/day more (p = 0.0001)
· The extra came from carbohydrate and fat; protein was the same in both periods

Weight change (2 weeks):

· UPF arm: +0.9 kg
· Unprocessed arm: −0.9 kg
· Net difference 1.8 kg / 2 weeks ≈ 4 lb

Eating rate:

· The UPF arm ate about 17 kcal (about 7.4 g) more per minute than the unprocessed arm
· p < 0.0001

Hormonal response:

· Compared with the UPF weeks, the appetite-suppressing hormone PYY was higher after meals in the unprocessed weeks
· Compared with admission, the hunger hormone ghrelin was lower in the unprocessed weeks
· In the UPF weeks neither changed this way

Energy expenditure:

· Total daily energy expenditure did not differ significantly (slightly higher in the UPF weeks, not statistically significant)
· So the extra weight came from eating more, not burning less

What it shows:

· Energy balance still holds: weight followed intake
· What changed was how much people ate: with sugar, fat, sodium and fibre matched, industrially restructured food still led to about 500 kcal a day more
· Only 20 people, 2 weeks per diet — it speaks to short-term intake, not long-term outcomes

3 · why

The trial did not isolate why people ate more on matched diets; five candidates are energy density, eating rate, food matrix, hyper-palatability and the microbiome with additives.Why eat more when everything was matched? The trial itself measured how much and how fast people ate and a few hormones; it did not isolate a cause. Below are 5 candidate explanations that do not exclude each other; the first two are the authors' own:

① Energy density (kcal/g):

· Counting beverages, the three daily meals as served were matched for energy density (1.02 vs 1.03 kcal/g)
· The non-beverage food: arm 1.96 kcal/g, unprocessed arm 1.06 kcal/g, about 85% higher
· The same weight of food carries far more calories on UPF; the authors think this gap likely contributed to the extra intake
· Fullness tracks, to a large extent, the amount of food eaten; the same mouthful of a denser food carries more calories

② Eating rate:

· About 17 kcal/min faster on the UPF diet
· Fullness signals take a while to catch up
· Eat fast, and more of the meal is down before they arrive
· UPF industrial extrusion + high water + soft texture → less chewing + faster swallowing

③ Food matrix destruction (not specifically tested in this trial):

· In whole foods, fibre + protein + fat + carb are structurally bound
· → slow gastric emptying + flat glucose response
· UPF dismantles + reassembles these: the same carb in flour + sugar vs whole-grain bread has can differ in gastric emptying + glucose response
· Classic contrast: whole apple vs apple juice vs apple-flavoured gummies

④ Hyper-palatability:

· Precisely engineered sugar + fat + salt + flavour combinations
· Some researchers think it repeatedly activates reward circuits, resembling a behavioural addiction — a hypothesis. Schulte 2015 used the Yale Food Addiction Scale in a survey and found highly processed foods most often linked with addictive-like eating; it was a questionnaire study, not a brain measurement
· The food industry's so-called bliss point: ratios tuned so that people keep eating

⑤ Microbiome + additives:

· Industrial emulsifiers disrupt gut bacteria + inflammation + metabolic syndrome in mice (Chassaing 2015 Nature tested CMC and polysorbate-80 — not carrageenan)
· Human evidence still accumulating

Special position of sugar-sweetened beverages:

· Liquid calories are compensated for less than the same calories eaten as solid food: in DiMeglio 2000, 15 adults took the same daily load of sugar; as jelly beans, they ate less of everything else, as soda they barely did, and weight rose only in the soda period (a small crossover trial)
· A 350 ml can of soda is about 150 kcal, and dinner does not shrink to match, so most of it is extra
· If you can change only one thing first, cutting sugary drinks is the easiest win

4 · paradigm shift

With sugar, fat, sodium and fibre matched people still ate more, which suggests processing itself matters; for long-term outcomes there are only observational studies.The shift Hall 2019 brought: earlier advice focused on single nutrients — less sugar, less fat, less sodium, count calories. This trial matched sugar, fat, sodium and fibre, and people still ate about 500 kcal a day more, which suggests that how food has been processed matters in itself. It had only 20 people on each diet for 2 weeks, so it speaks to short-term intake, not long-term disease.

For long-term outcomes there are only observational studies: the Lane 2024 (BMJ) umbrella review pooled 45 results from 14 covering about 9.8 million people; 32 of them (71%) linked higher ultra-processed intake with worse outcomes:

· Type 2 diabetes: each 10% more of total energy from ultra-processed food, 1.12; moderate certainty, the most certain finding
· Cardiovascular death: highest vs lowest intake, risk ratio 1.50; very low certainty
· All-cause death: highest vs lowest intake, risk ratio 1.21; low certainty
· Depression: risk ratio 1.22 (pooled observational studies, not a randomized trial)
· Cancer: in the French NutriNet-Santé cohort, each 10% more ultra-processed food came with an overall cancer of 1.12 (Fiolet 2018)

These figures show a consistent direction, not precise risks; observational studies cannot remove every .

In practice:

· Swap sugary drinks and bubble tea first: water, unsweetened tea, black coffee
· Fewer packaged snacks: crisps, biscuits, sweets, factory pastries → nuts, fruit, plain yogurt
· Less processed meat: sausages, bacon, ham, luncheon meat → fresh meat, fish, eggs, beans; processed meat is an IARC Group 1 carcinogen (Group 1 describes certainty of evidence, not size of risk)
· No need to go to zero: birthday cake or the occasional bubble tea is fine; the aim is for whole foods to make up most of what you eat

So look not only at which nutrient, but also at how far the food itself has been industrially restructured.

Full mechanism, in text

1 · design

2 · result

3 · why

4 · paradigm shift

UPF · Hall 2019 NIH RCT · nutrients matched, still ate more

Full mechanism, in text

1 · design

2 · result

3 · why

4 · paradigm shift

UPF · Hall 2019 NIH RCT · nutrients matched, still ate more