Chapter 22 — Further Reading

Real sources only. Where I'm unsure of exact details, I describe the source rather than inventing a citation. ⚠️ This page closes Part IV, so the last section is about the whole part.

⚠️ Read the ward trial. Actually read it.

Hall KD et al., "Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain," Cell Metabolism, 2019.

⚠️ This is the single most important primary source in this book and it is readable by a non- specialist. Read it in this order:

1. The seven-day menus in the supplementary materials. ⚠️ This is the part almost nobody reads and it is the most instructive. Seeing what "ultra-processed" and "unprocessed" actually meant on a plate does more than any definition.

2. The matching table. ⚠️ Check for yourself what was and wasn't matched — and note the fiber supplementation in the UPF arm's beverages, which is a real design choice with real implications.

3. Then the results. Then the limitations section, which is candid.

And then read Hall's subsequent work, searching "Hall ultra-processed energy density hyperpalatable controlled feeding." ⚠️ He has published findings that complicate his own headline result and said so publicly. Watch what that looks like — it is rare in this field and it is the best available demonstration of Chapter 1 §1.5's test being passed.

On NOVA itself

Monteiro CA et al. — search "NOVA food classification Monteiro" for the primary papers, and the FAO's 2019 report on ultra-processed foods (fao.org), which is free and sets out the framework at length.

⚠️ Read the classification's own definitions before reading anyone's argument about it. You will find it is more careful than either its enthusiasts or its critics represent.

Then read the critiques, properly:

Mike Gibney's published critiques of NOVA — search "Gibney ultra-processed food critique." ⚠️ He is a serious nutrition scientist making serious methodological objections, particularly about classification reliability and actionability. §22.7 is built on this literature and it deserves to be read directly rather than through me.

On reliability specifically: search "NOVA classification inter-rater reliability agreement." ⚠️ This is the objection I find hardest to answer and I'd rather you saw the data yourself.

On the mechanisms

On energy density: Barbara Rolls's body of work — search "energy density satiety Rolls". ⚠️ This is decades of careful experimental work and it substantially predates the UPF framing.

On eating rate: search "eating rate energy intake oral processing" and "food texture eating rate satiety." ⚠️ The experimental work here is elegant and underappreciated — and it's the mechanism the ward trial measured directly.

On hyper-palatability: search "hyperpalatable foods definition Fazzino" for the attempts to define the concept operationally. ⚠️ Note that the definitions are still being worked out, which is why §22.5 rates it 🟡 and not 🟢.

On the cohort evidence

NutriNet-Santé and the other large cohorts — search "ultra-processed food consumption mortality cohort" and "ultra-processed food cardiovascular disease prospective."

⚠️ Read at least one full paper including its limitations section, and note how consistently the authors flag confounding by socioeconomic position, smoking and overall diet quality. They are more careful than the coverage of them.

Then read a sceptical review, searching "ultra-processed food epidemiology confounding critique." ⚠️ Chapter 20 §20.7's healthy-user problem runs in the opposite direction here and is just as strong.

On additives and emulsifiers

Chassaing B, Gewirtz AT et al. — the rodent work on carboxymethylcellulose and polysorbate-80, and then ⚠️ the small human randomized controlled feeding trial of CMC (Gastroenterology, 2022).

⚠️ Read the human study's sample size and duration before its conclusions. This is genuinely interesting early work and it is being cited as though it were settled.

Chapter 27 goes into the microbiome properly. ⚠️ If you find this area exciting, read Chapter 27 before you act on anything in it.

On the food environment and equity

On cost: search "cost per calorie ultra-processed versus minimally processed food" and "diet cost socioeconomic gradient." ⚠️ §22.11's claim is not rhetorical and the data is unambiguous.

On time and cooking: search "time poverty cooking from scratch food preparation time" — the sociological literature here is better than the nutrition literature and it is rarely cited by nutritionists.

Chapter 31 (meal planning), Chapter 32 (economics) and Chapter 36 (food systems) take this up properly. ⚠️ If §22.11 was the section that landed hardest for you, read Chapter 32 next rather than Chapter 23.

On policy

On the salt reduction programmes — search "UK salt reduction programme evaluation". ⚠️ Read this as the model of a nutrient rule that worked, and work out why (gradual, category-wide, monitored, palates adapted together). Case Study 2's one success.

On front-of-pack labelling: Chile's warning-label regulations are the most studied recent example — search "Chile front-of-pack warning labels evaluation." ⚠️ Note that Chile's system regulates nutrient thresholds, not formulation — which is §22.10b's tension, in force.

On whether UPF can be regulated at all: search "ultra-processed food policy regulation feasibility." ⚠️ The honest answer in the literature is that nobody has solved §22.7's fourth objection.

⚠️ A note on what to be careful with — and this one is about Part IV as a whole

This chapter's specific risk is that it becomes the sixth villain.

Three failure modes, and I'd guard against all of them:

1. ⚠️ Moral inflation. "Ultra-processed" describes how a food was made, not the person eating it. Chapter 20 §20.13's clean-eating structure is fully available here — a binary, morally-loaded category with no defined endpoint — and it is the most likely way this framework causes harm. Chapter 34.

2. ⚠️ Over-reading the trial. It established an intake effect over two weeks in twenty people. It did not establish that ultra-processed food causes disease independently of the energy it delivers, and that distinction is the difference between this chapter and the movement it describes. §22.14 holds that claim at LOW confidence and I'd want you to hold it there too.

3. ⚠️ Ignoring §22.7. The reliability problem is real. The edge cases are genuinely awkward. A better classification will probably replace NOVA. If you take the framework and skip the objections, you have done exactly what Part IV's five factions did with their own evidence.

And the closing thought for Part IV

⚠️ Six chapters, and the practical column of §22's closing table contains almost nothing anyone argues about.

Drink less soft drink. Eat more vegetables. Eat fish twice a week. Wash your hands. Don't reuse frying oil. Take folic acid if you might get pregnant. Eat less ultra-processed food, in whatever way your circumstances permit.

⚠️ That is the yield of the loudest arguments in modern nutrition, and it was available in Chapter 10.

What the six chapters actually bought you is not the list. It's the method — the five slots, the numbered claims, the discriminating design, the documented-episode test, the confidence table, and the habit of asking what a claim is being paid to do.

⚠️ The list will change. Some of it will be wrong within a decade (Chapter 17's three-year reminder). The method is what you keep.

Part V is where the arguing stops and the applying starts.