Cut ultra-processed food

TL;DR. Ultra-processed food is the single biggest dietary lever on chronic disease. The category is called NOVA Group 4. Ingredients set it, not nutrients. Any modified starch, emulsifier (polysorbate 80, DATEM, mono- and diglycerides), hydrolyzed protein, maltodextrin, or industrial gum (xanthan, carrageenan, guar) puts a product in. A loud health claim plus a long ingredient list usually puts it in too. Push your UPF share from the U.S. and U.K. average of 50 to 60 percent down toward 30 percent. Swap, don't quit.

What this goal does

This goal flags packaged products as NOVA Group 4 and points you to NOVA 1 to 3 swaps in the same role. A flag is information, not a verdict. The scanner shows you the failing ingredient and the closest real-food version of the same thing. This page covers the rules. The science lives in the deep dive, The ultra-processed food problem.

Evidence in three paragraphs

Carlos Monteiro published NOVA in 2009. He sorted food by how it gets made, not by its nutrients. Group 1 is whole food. Group 2 is cooking ingredients. Group 3 is real food preserved with Group 2 ingredients using methods a home cook knows. Group 4 is built in a factory from extracted substances and additives no home cook owns. Around 250 large studies now track UPF share against outcomes. Every 10 percent jump in UPF intake raises all-cause death about 15 percent, type 2 diabetes about 15 percent, and dementia about 25 percent. The effect holds after adjusting for nutrients, weight, smoking, income, and exercise.

Kevin Hall ran the cleanest cause-and-effect test at the NIH in 2019, in Cell Metabolism. Twenty adults lived on the metabolic ward for four weeks. Each ate an 80 percent UPF diet for two weeks and an 80 percent NOVA 1 to 3 diet for two weeks. The diets matched on calories offered, sugar, salt, fat, and fiber. On the UPF arm, the same people ate around 500 extra calories a day and gained about a pound a week. Hunger scores matched. Tim Spector replicated the design in 2023 with identical-twin pairs for BBC Panorama. Inflammation, postprandial glucose, and blood-fat markers split between the twins inside two weeks. The "n equals 20" objection is settled.

Chris van Tulleken names four engineering levers behind the overeating. One, a destroyed food matrix that makes food soft, dry, and calorie-dense. Two, salt, sugar, and fat stacked at the bliss point. Three, taste signals that do not match what your body gets, like low-calorie sweeteners. Four, gut damage from emulsifiers like polysorbate 80 and carboxymethylcellulose. The UK's Scientific Advisory Committee on Nutrition declined in July 2023 to act on UPF. Children in the U.S. and U.K. now get roughly 75 percent of their calories from UPF; infants under two get about 69 percent. The 30 percent adult target is reasonable; the kid number is an emergency.

What helps

• Swap the single UPF product you eat most days for a NOVA 1 to 3 version of the same role. One swap moves your share more than a week of effort spread across many products.
• Cut sweet drinks first. Soda, energy drinks, sweetened plant milks, sports drinks, vitamin water, and sweet coffee drinks. Water, plain coffee, plain tea, or plain sparkling water do the job.
• Keep frozen vegetables, tinned beans, tinned tomatoes, sardines, plain whole-milk yogurt, real cheese, real bread, dried pasta, eggs, fruit, oils, and vinegar in the house. These are NOVA 1 to 3 and let you build meals without a recipe.
• Read the ingredient list. The front of the package is marketing.

What hurts

• Trying to quit UPF cold. Hall's data says lower share, not zero. People who chase zero rebound to the same share by month three.
• Trusting front-of-pack health claims. As van Tulleken puts it, almost every food with a health claim on the packet is UPF. The claim itself is a tell.
• Treating "lower sugar" or "added protein" as progress. Reformulation keeps the product in Group 4. The emulsifiers, modified starches, and flavorings stay.
• Swapping one UPF for another. A Beyond Burger is not a step down from a Quest Bar. Both are extreme Group 4.

How the scanner uses this

Hard rules. Any one of these in the ingredient list flips a product to NOVA Group 4 on its own:

• Emulsifiers: polysorbate 60 or 80, DATEM, mono- and diglycerides, carboxymethylcellulose, gellan gum.
• Modified starches of any kind: modified corn starch, modified wheat starch, modified food starch, hydroxypropyl distarch phosphate.
• Hydrolyzed protein isolates: soy protein isolate, pea protein isolate, whey protein isolate, hydrolyzed wheat protein, hydrolyzed vegetable protein.
• Maltodextrin and other industrial carbohydrates: glucose syrup, dextrose, "corn syrup solids."
• Gums used as texture builders: xanthan, carrageenan, guar, locust bean.

Soft rules. A product is flagged Group 4 when both apply:

• A front-of-pack health claim ("high protein," "good source of fiber," "made with real fruit," "no added sugar," "plant-based," "natural").
• Six or more ingredients on the back.

Van Tulleken's pattern: a health claim on a long ingredient list is almost always UPF. The soft rule catches dressed-up Group 4.

Bonuses. A product earns a positive flag when one of these applies:

• Single-ingredient or NOVA Group 1 to 2 (rolled oats; whole apple; almonds; olive oil; plain yogurt).
• An ingredient list a home cook could replicate (sourdough: flour, water, salt, starter; marinara: tomatoes, olive oil, garlic, basil, onion, salt).

Worked examples

• Cliff Bar. Group 4. Soy protein isolate, brown rice syrup, "natural flavors." Hard rule: protein isolate. Soft rule also fires.
• Lunchables. Group 4. Mechanically separated meat, modified food starch, sodium phosphates, soy protein concentrate. Worst-case Group 4.
• Beyond Burger. Group 4. Pea protein isolate, methylcellulose, modified food starch, "natural flavors." Extreme Group 4 dressed as a health product.
• Quest Bar. Group 4. Whey and milk protein isolates, soluble corn fiber, erythritol, sucralose. Two isolates plus a sweetener.
• Lara Bar (original flavors). Group 1 to 2. Dates, nuts, sometimes a spice. Passes. Newer "protein" variants add pea or whey isolate and flip to Group 4.
• Plain Greek yogurt. Group 1. Milk and live cultures.
• Whole apple. Group 1. Single ingredient.

NOVA Group 4 is a wide tent. A Lunchable, a Quest Bar, and a Beyond Burger all fit in it. The category does not mean "junk." It means "built from industrial ingredients a home cook does not own." A loud health claim on the front does not protect a product; the claim is the tell.

Related reading

• The ultra-processed food problem. The full deep dive on NOVA, Hall's trial, the four engineering levers, the dose-response epidemiology, and why reformulation does not work.
• Real food vs. edible foodlike substance. Pollan's pocket version of the same rule, with the 30-second ingredient-list test.
• Grocery shopping in 15 minutes. The store-level application: perimeter-first route plus the 5-second label scan.

Sources

• Monteiro, C. A. (2009). Nutrition and health. The issue is not food, nor nutrients, so much as processing. Public Health Nutrition, 12(5), 729 to 731. DOI: 10.1017/S1368980009005291.
• Hall, K. D., Ayuketah, A., Brychta, R., et al. (2019). Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake. Cell Metabolism, 30(1), 67 to 77. DOI: 10.1016/j.cmet.2019.05.008.
• Chassaing, B., Koren, O., Goodrich, J. K., et al. (2015). Dietary emulsifiers impact the mouse gut microbiota promoting colitis and metabolic syndrome. Nature, 519, 92 to 96. DOI: 10.1038/nature14232.
• van Tulleken, C. (2023). Ultra-Processed People: The Science Behind the Food That Isn't Food. Knopf. NOVA Group 4 operational test, four engineering levers, health-claim-as-UPF-tell, summary of ~250 dose-dependent epidemiology papers. Chapter 2 in particular.
• Lustig, R. H. (2021). Metabolical: The Lure and the Lies of Processed Food, Nutrition, and Modern Medicine. Harper Wave. UPF as absence of fiber plus added sugar plus industrial substances; protect-the-liver, feed-the-gut frame.
• Spector, T. (2022). Food for Life. Jonathan Cape. 2023 BBC Panorama TwinsUK replication of Hall 2019; ~75% paediatric UPF share; UK SACN July 2023 inaction. Chapters 11 and 33.