For the first time in U.S. history, federal dietary guidelines explicitly name ultra-processed foods as a foundational nutritional problem rather than a peripheral concern.
The 2025-2030 Dietary Guidelines call for "dramatic reduction" in consumption of "highly processed foods laden with refined carbohydrates, added sugars, excess sodium, unhealthy fats, and chemical additives." While the guidelines don't use the specific term "ultra-processed," the description maps directly onto the NOVA classification system—a framework developed by Brazilian researchers that categorizes foods by degree and purpose of processing.
This represents a significant policy shift. Previous guidelines focused on individual nutrients (saturated fat, sodium, sugar) without addressing how industrial food processing itself might affect health outcomes independent of nutrient composition. The new guidance acknowledges what epidemiological research has been showing for years: the degree of processing matters, not just the nutrient profile.
But most Americans—and many health professionals—don't actually understand what "ultra-processed" means. The term gets used as shorthand for "bad food," which isn't scientifically precise and creates confusion about what people should actually avoid.
If ultra-processed foods are now officially recognized as problematic, we need clarity on what that category actually includes, what the evidence shows, where the classification system works well, and where it has limitations.
What Is the NOVA Classification?
NOVA divides all foods into four groups based on the extent and purpose of processing:
Group 1: Unprocessed or Minimally Processed Foods Foods altered by processes like drying, freezing, pasteurization, or fermentation that don't add substances like salt, sugar, oils, or preservatives.
Examples: Fresh fruits and vegetables, dried beans, frozen vegetables (plain, no sauce), fresh/frozen meat and fish, plain milk, plain yogurt, eggs, rice, oats, nuts, seeds, fresh herbs and spices, coffee, tea.
Processing purpose: Preservation, preparation for eating, or minimal transformation.
Group 2: Processed Culinary Ingredients Substances extracted from Group 1 foods or from nature through processes like pressing, refining, grinding, or milling. Used in cooking, not consumed alone.
Examples: Vegetable oils, butter, lard, sugar, honey, salt, starches, vinegar.
Processing purpose: Used to prepare and season Group 1 foods.
Group 3: Processed Foods Foods made by adding Group 2 ingredients (salt, sugar, oil) to Group 1 foods, using preservation methods like canning or bottling.
Examples: Canned vegetables with added salt, canned fish, cheese, freshly made bread (flour + water + salt + yeast), salted/sugared nuts, cured/smoked meats.
Processing purpose: Increase durability and modify/enhance sensory qualities.
Group 4: Ultra-Processed Foods Industrial formulations made predominantly from substances extracted from foods (oils, fats, sugar, starch, protein isolates) or synthesized from food constituents (hydrogenated fats, modified starches) or synthesized in labs (flavor enhancers, colors, emulsifiers, humectants, sweeteners, thickeners).
Typically contain five or more ingredients, many with names you wouldn't recognize as food. Often designed to be hyper-palatable, convenient, and shelf-stable.
Examples: Sodas, packaged snacks (chips, cookies, candy), ice cream, mass-produced packaged breads, breakfast cereals, energy bars, instant soups and noodles, chicken nuggets, hot dogs, many plant-based meat alternatives, protein powders, meal replacement shakes, most items in the center aisles of grocery stores.
Processing purpose: Create branded, convenient, hyper-palatable products for mass distribution and profit.
The key distinction: Groups 1-3 involve processing that preserves, prepares, or enhances actual foods. Group 4 involves industrial formulation of substances into edible products designed for convenience and profitability more than nutrition.
The Evidence Linking Ultra-Processing to Health Outcomes
The research connecting ultra-processed food consumption to adverse health outcomes is substantial and growing:
Cardiovascular Disease: Multiple large cohort studies (NutriNet-Santé in France, UK Biobank, Nurses' Health Study) show associations between high UPF consumption and increased risk of cardiovascular disease, hypertension, and cardiovascular mortality. A 2019 BMJ study of 105,000 adults found that a 10% increase in ultra-processed food proportion of diet was associated with 12% higher cardiovascular disease risk.
Metabolic Health: Higher UPF consumption correlates with increased obesity, type 2 diabetes, and metabolic syndrome across multiple populations. A 2020 study in JAMA Internal Medicine following 20,000+ adults found those in the highest quartile of UPF consumption had 15% higher risk of type 2 diabetes.
Cancer: The NutriNet-Santé cohort (105,000 participants) found 10% increase in UPF consumption associated with 12% higher overall cancer risk and 11% higher breast cancer risk. Mechanisms aren't fully understood but likely involve food additives, packaging contaminants, and nutrient displacement.
All-Cause Mortality: Multiple studies show associations between high UPF intake and increased all-cause mortality. A 2019 BMJ study found those in highest quartile of UPF consumption had 62% higher risk of all-cause mortality compared to lowest quartile.
The NIH Controlled Feeding Study: Perhaps most compelling is Kevin Hall's 2019 randomized controlled trial—the gold standard of evidence. Participants were provided either ultra-processed or unprocessed diets matched for calories, sugar, fat, fiber, and macronutrients. When eating ultra-processed foods, participants consumed ~500 more calories per day and gained weight. On unprocessed foods, they ate less and lost weight. Same nutrient composition. Different outcomes based on processing alone.
This suggests ultra-processing affects satiety, eating rate, or energy absorption in ways that nutrient analysis doesn't capture.
Important caveat: Most evidence is observational (correlation, not proven causation). People who eat more ultra-processed foods tend to have other risk factors—lower income, less education, more sedentary behavior. Researchers adjust for these confounders, but residual confounding always remains possible. The Hall study provides causal evidence, but it's one trial over four weeks—we need more long-term RCTs.
Where NOVA Works Well
1. Captures what nutrient-based analysis misses Two foods can have identical macronutrients but wildly different effects on satiety, eating behavior, and health. NOVA acknowledges this by focusing on food structure and industrial formulation, not just nutrient composition.
Example: Homemade oat cookies (Group 3: oats + butter + sugar + eggs) vs. packaged sandwich cookies (Group 4: modified starches + hydrogenated oils + high-fructose corn syrup + emulsifiers + artificial flavors). Similar calories and macros. Different food matrices, different satiety, different metabolic effects.
2. Identifies displacement of whole foods Ultra-processed foods tend to be energy-dense but nutrient-poor. High UPF consumption correlates with lower intake of fiber, vitamins, minerals, and phytonutrients—not because UPF actively harms, but because it displaces foods that provide essential nutrients.
3. Correlates with health outcomes across diverse populations The UPF-health association appears in multiple countries with different food systems, suggesting the pattern is robust, not an artifact of specific cultural contexts.
4. Provides actionable public health messaging "Reduce ultra-processed foods" is clearer than "optimize saturated fat to polyunsaturated fat ratio" or "manage omega-6 to omega-3 balance." NOVA translates complex nutritional science into practical guidance.
5. Highlights the role of industrial food systems NOVA makes visible the fact that most grocery store products aren't "food"—they're industrially formulated edible products designed for profit maximization, not nutritional value. This shifts focus from individual blame to systemic critique.
Where NOVA Has Limitations
1. The boundaries are fuzzy Is store-bought whole grain bread ultra-processed (contains emulsifiers, dough conditioners) or processed (mostly flour + water + yeast + salt)? Different NOVA researchers classify it differently. Is plain Greek yogurt with added probiotics ultra-processed (bacterial cultures added) or minimally processed (just milk fermentation)? The classification requires judgment calls.
2. Not all Group 4 foods are equally harmful Plain protein powder (whey isolate + sweetener) is technically ultra-processed. So is a high-fiber cereal with added vitamins. So is a candy bar. NOVA treats these identically, but they have different nutritional and health implications.
Some ultra-processed foods might actually improve nutrition access for specific populations: fortified foods for people with micronutrient deficiencies, protein supplements for older adults with inadequate intake, meal replacement shakes for people with chewing difficulties.
3. Doesn't account for portion size or frequency Eating ultra-processed foods occasionally in small amounts likely has negligible health impact. But NOVA classifications are binary—a food either is or isn't ultra-processed, with no gradation for context.
4. Can create nutritional oversimplification Focusing exclusively on processing ignores other important factors: Does the food provide adequate protein? Fiber? Essential micronutrients? What's the glycemic impact? The energy density?
A minimally processed food isn't automatically nutritionally optimal (white rice is Group 1 but nutrient-poor with high glycemic impact). An ultra-processed food isn't automatically devoid of value (fortified cereals provide iron and B vitamins many people lack).
5. May be culturally biased NOVA emerged from Brazilian food culture and public health context. Some traditional preservation methods common in other cultures (fermented soy products, certain pickled vegetables, dried fish products) don't fit neatly into the framework.
6. Provides limited guidance for individual optimization Someone managing diabetes needs different nutritional priorities than someone training for endurance sports than someone managing kidney disease. Processing level matters, but it's not the only thing that matters—and for certain conditions, it might not be the primary thing that matters.
Processing as One Dimension Among Many
This is where NOVA fits into the multi-dimensional evaluation framework I've been developing:
Processing (NOVA classification) measures: Industrial formulation, food structure, ingredient complexity, and manufacturing methods.
But it doesn't measure:
- Glycemic impact (how it affects blood sugar)
- Nutrient density (vitamins, minerals, fiber per calorie)
- Energy density (calories per gram, satiety properties)
- Protein adequacy (muscle maintenance, satiety, metabolic support)
- Cardiovascular markers (sodium, saturated fat, potassium, fiber)
- Digestive tolerance (FODMAP content, fiber type, probiotic presence)
A food can be:
- Minimally processed + nutrient-poor + high glycemic: White rice
- Ultra-processed + relatively nutrient-dense + high protein: Protein bar with added vitamins
- Processed + nutrient-dense + low glycemic: Canned beans with added salt
- Minimally processed + nutrient-dense + low glycemic + high satiety: Salmon
Processing level provides valuable information. But it's one data point among many.
For someone managing type 2 diabetes, glycemic impact might matter more than processing. For someone with severe IBS, digestive tolerance might matter more than NOVA group. For an older adult with inadequate protein intake, protein density might matter more than whether the source is ultra-processed.
This doesn't mean processing doesn't matter. It means processing interacts with other nutritional properties and individual health contexts to determine optimal choices.
How This Works in Practice
In the Focus-Fit Health Scoring system, NOVA classification is integrated into several dimension scores but isn't the sole determinant of any:
Glycemic Impact Score (GIS): Ultra-processing (NOVA 4) receives a penalty because industrial processing often increases glycemic response—but the penalty is combined with carbohydrate-to-fiber ratio and sugar content. A minimally processed food with high sugar (fruit juice) still scores lower than whole fruit. An ultra-processed food with moderate carbs and added fiber might score higher than white bread.
Nutrient Density: Ultra-processing correlates with nutrient dilution, so NOVA 4 foods receive penalties—but actual micronutrient content matters more. A fortified ultra-processed cereal might score higher than white rice despite being more processed.
Cardiovascular Health: Ultra-processing often correlates with high sodium and industrial fats, affecting the cardiovascular dimension—but actual sodium, saturated fat, fiber, and potassium content determine the score.
The principle: Processing is one input among many. The scores measure physiological impact, not moral judgment about food purity.
The Practical Minimum for Most People
Based on the evidence, the clearest nutritional intervention is:
Substantially reduce Group 4 ultra-processed foods. Focus most consumption on Groups 1-3.
This doesn't mean perfection. It means:
- Most meals built around whole or minimally processed foods (Group 1)
- Using processed culinary ingredients for cooking (Group 2: oils, butter, salt, sugar in reasonable amounts)
- Including some processed foods for convenience (Group 3: canned beans, plain yogurt, cheese, whole grain bread)
- Limiting ultra-processed foods to occasional consumption rather than dietary foundation
For most people in most contexts, this shift alone—without obsessing over macronutrient ratios, micronutrient optimization, or meal timing—resolves a substantial portion of diet-related metabolic dysfunction.
But context still matters:
Someone with diabetes might need to prioritize low-glycemic foods within Groups 1-3 rather than assuming all minimally processed foods are equally appropriate.
Someone with limited budget might need to strategically use some Group 4 foods (fortified cereals, protein powders, shelf-stable options) when access to affordable Group 1-3 alternatives is limited.
Someone with digestive conditions might need to modify fiber content within Group 1 foods rather than assuming all minimally processed foods are well-tolerated.
Processing level provides the foundation. Individual context determines the specifics.
What Still Needs Research
While the association between ultra-processed foods and adverse health outcomes is well-established observationally, several questions remain:
1. Mechanisms: Why do ultra-processed foods promote overconsumption and poor health? Is it:
- Food additives and emulsifiers disrupting gut microbiome?
- Faster eating rate and reduced satiety from altered food structure?
- Displacement of fiber and essential nutrients?
- Packaging contaminants migrating into food?
- Some combination?
Understanding mechanisms would help identify which aspects of ultra-processing matter most and inform reformulation efforts.
2. Dose-response: What threshold of ultra-processed food consumption shows measurable health impact? Is it linear (every reduction helps) or threshold-based (need substantial reduction to see benefit)?
3. Individual variation: Do some people tolerate ultra-processed foods better than others based on genetics, microbiome, or metabolic health status?
4. Reformulation potential: Can industrial food processing be modified to reduce harm while maintaining affordability and convenience? Or is ultra-processing inherently problematic regardless of specific formulation?
5. Long-term intervention trials: The Hall study showed short-term effects. We need multi-year RCTs examining whether sustained UPF reduction improves biomarkers and clinical outcomes.
My Position: NOVA Is Valuable But Insufficient
The evidence supporting the ultra-processed food hypothesis is compelling. The new guidelines' explicit recognition of processing as a nutritional concern reflects genuine scientific consensus.
NOVA provides valuable information:
- It captures aspects of food that nutrient analysis misses
- It correlates with health outcomes across diverse populations
- It provides actionable public health messaging
- It focuses attention on industrial food systems
But NOVA alone is insufficient for individual nutritional optimization:
- It doesn't measure glycemic impact, nutrient density, or protein adequacy
- It treats all Group 4 foods identically despite different nutritional profiles
- It doesn't account for individual health contexts requiring different priorities
The optimal approach: Use NOVA as one dimension within multi-dimensional evaluation. Processing level matters—and so do glycemic impact, nutrient density, protein adequacy, cardiovascular markers, and digestive tolerance.
Measure everything independently. Let individual context determine how to weight the dimensions.
What I'm Seeking
If you're a researcher interested in studying:
- Mechanisms linking ultra-processing to health outcomes
- Dose-response relationships for UPF consumption
- Individual variation in metabolic response to ultra-processed foods
- Integration of NOVA classification into multi-dimensional nutrition assessment
- Implementation research on reducing UPF consumption in diverse populations
I'd welcome conversations about collaborative research.
I have operational infrastructure that integrates NOVA classification with other nutritional dimensions. I have commercial interests. I also have genuine interest in understanding where processing matters most and for whom—and that requires rigorous research, not assumptions.
Disclosure: I am the founder and CEO of Digital Galactica Labs, which develops nutrition intelligence systems that incorporate NOVA classification alongside other nutritional dimensions. I have filed nine provisional patent applications covering these systems and have commercial interests in their development. I am not a clinician, registered dietitian, or licensed healthcare provider. The systems described are operational research infrastructure, not validated clinical interventions. Independent validation through controlled trials is explicitly needed and welcomed.
