Extremely high priority
18Nice to haveEvidence: solid
Minimize ultra-processed food
In the only tightly controlled feeding trial, adults offered an ultra-processed diet passively ate ~500 kcal/day more and gained 0.9 kg in two weeks, and an 8-week free-living RCT found roughly double the weight loss on a minimally processed diet — that is the causal core, and it covers intake and short-term weight only. The larger claims (mortality +21%, diabetes +12%, depression) come from cohorts and are associations, not proof. Start with the worst offenders — sugary drinks, processed meats, packaged snacks — and shift toward short-ingredient-list foods, mostly by cooking more of your own meals.
Priority Matrix
The numbers.
Four dimensions, scored 1–6, where six is always the good end: cheapest, quickest, easiest, highest impact. Priority is Impact squared, divided by what it costs you in energy, money and time.
CostGroceries run ~40% more per calorie
TimeShared with cooking at home
EnergySustained swaps vs cheap defaults
ImpactRCT: intake and short-term weight
How sure we are
SolidGood trials pointing the same way.
Kept separate from the priority score on purpose: confidence in the evidence never raises how worth-it something is.
What we read
- Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain
Cell Metabolism (Hall et al.) · 2019
Inpatient randomized crossover, n=20: +508 kcal/day and +0.9 kg in 2 weeks on the UPF arm despite matched presented calories, macros, sugar, sodium, and fiber; UPF meals eaten ~50% faster; unprocessed ingredients cost ~40% more.
- UPDATE trial: minimally processed vs ultra-processed diets matched to UK dietary guidance
Nature Medicine (Dicken et al.) · 2025
8-week free-living crossover, n=50: −2.06% body weight on the minimally processed arm vs −1.05% on the UPF arm, both within guideline-compliant diets; one trial, with published critiques in correspondence.
- Ultra-processed food exposure and adverse health outcomes: umbrella review
BMJ (Lane et al.) · 2024
45 meta-analyses, ~9.9M participants: convincing associations for CVD mortality (~+50%), anxiety/common mental disorders (+48–53%), type 2 diabetes (+12%); highly suggestive for all-cause mortality (+21%) and depression (+22%) — all observational, GRADE certainty mostly low/very low.
- Ultra-processed foods and risk of all-cause mortality: an updated systematic review and dose-response meta-analysis
Systematic Reviews · 2025
18 cohorts, 1.15M participants: each 10% of energy from UPF associated with ~10% higher all-cause mortality (HR 1.10, 1.04–1.16); association only — no trial has ever measured mortality.
- Does the concept of ultra-processed foods help inform dietary guidelines? (Great Debates)
American Journal of Clinical Nutrition (Astrup & Monteiro) · 2022
NOVA group 4 spans consistently-harmful soda and processed meat to neutral-or-protective wholegrain breads and cereals; both sides agree processing matters, disagree on whether NOVA adds information beyond nutrient content.
- Nut consumption and risk of cardiovascular disease, cancer, and mortality: dose-response meta-analysis
BMC Medicine (Aune et al.) · 2016
Per 28 g/day: CHD RR 0.71 (0.63–0.80), CVD 0.79, all-cause mortality 0.78, total cancer 0.85; stroke null — the independent evidence base for nuts as a UPF-displacing snack.
- Primary Prevention of Cardiovascular Disease with a Mediterranean Diet Supplemented with Extra-Virgin Olive Oil or Nuts (republished)
New England Journal of Medicine (Estruch et al.) · 2018
PREDIMED nuts arm: HR 0.72 (0.54–0.95) for major cardiovascular events in the 2018 republished analysis, after the 2013 original was retracted over randomization irregularities.
- Can Chocolate Be Classified as an Ultra-Processed Food?
PMC (Bortoletto et al.) · 2023
Most commercial dark bars, including 85%, class as NOVA group 4 once they contain emulsifiers or flavors; only cocoa + cocoa butter + sugar bars class as group 3 — so chocolate cannot ride a UPF-avoidance card.
Research from NIH / National Institute of Diabetes and Digestive and Kidney Diseases · University College London · Deakin University Food & Mood Centre · Imperial College London. We cite their published work. They have no involvement with Anew.
Where this falls down
Costs more: the unprocessed arm of the NIH trial ran ~40% pricier per calorie, and this card taxes exactly what UPF is for — cheap, fast, shelf-stable food.
NOVA-4 is a blunt category: it lumps soda and processed meat (where the harm signal concentrates) with wholegrain packaged breads, fortified cereals, and yogurt drinks that look neutral-to-protective in subgroup analyses; cutting those first can lower fiber intake for zero expected gain.
Beyond short-term weight, every claimed benefit is observational with heavy confounding — high UPF intake tracks with lower income, smoking, and less exercise — and GRADE certainty in the umbrella review was low or very low for most outcomes.
The mechanism is unproven: energy density, eating rate, and hyperpalatability were confounded with processing in the feeding trial and may carry the whole effect.
All-or-nothing framing invites disordered restriction and social friction; the evidence supports shifting the share, not achieving zero.
Still unknown. Whether the harm is processing itself or the energy density, eating rate, and hyperpalatability that travel with it (Hall's follow-up work points to the latter, which would mean some reformulated UPF is fine); the dose threshold; which NOVA-4 subcategories beyond sugary drinks and processed meat actually matter; and no long-term randomized outcome data exist — probably ever.
Topics
Where it fits.
Related products
The gear.
Products that share a topic with this intervention. Same matrix scoring, same restraint.
- Important
Best ValueNutritionThe dessert with a cocoa-flavanol dossier.
Cost$4–8TimeNoneEnergyEffortlessImpactModerate
Keep going