1. The concept of UPF demonises well-formulated processed foods.
Oftentimes “ultra-processed food” is assumed to be a synonym for “junk food”. But this is a misconception. To understand why, let's look at some rough definitions:
- I define “junk foods” as convenience foods high in calories, refined fat and (/or) refined carbohydrates (e.g., sugar-sweetened beverages, biscuits, fast food, cakes, ready meals etc.).
- A food is identified as ultra-processed, if it contains “substances never or rarely used in kitchens, or classes of additives whose function is to make the final product palatable or more appealing (‘cosmetic additives')1. From what I can gather, the aforementioned substances and additives are taken to be signs of “industrial processing”.
Now, there is a considerable amount of overlap between these two concepts. It seems to me that most foods that contain “unusual substances” and/or cosmetic additives are junk foods. But importantly, many ultra-processed foods are not “junk”; that's to say, they are NOT hyperpalatable, energy dense convenience foods.
Here's a schematic representation of the point I am making:
The recommendation to avoid ultra-processed foods is not uncommon. If one erroneously assumes that “ultra-processed food” is another term for “junk food”, this may seem like incontestably good advice. But in reality, UPFs are foods that have been subjected to industrial processing irrespective of their nutrient profile. As such, many well-formulated processed foods fall under this umbrella term.
Note: if you're looking for a system that accurately and reliably identifies what might be called junk foods, check out the Nutrient Profile Model2.
2.The NOVA System of Food Classification is Illogical.
NOVA is by far and away the most widely used framework for classifying foods by their level of processing. Under NOVA foods are categorized into 4 groups:
Group 1 - Unprocessed or minimally processed foods
Group 2 - Processed culinary ingredients
Group 3 - Processed foods
Group 4 - Ultra-processed food & drink products
These groupings are meant to capture the extent to which a food has been processed. However, in practice the NOVA system often classifies foods in a disorganised way. I will attempt to show this with a series of examples:
Example 1: Isolated starches (e.g., cornstarch, potato starch) are classified as GROUP 2 (processed ingredients) and the fact that a food or drink product contains isolated starch does not entail that it's ultra-processed. On the other other, isolated proteins (e.g., soy protein, whey protein) are classified as GROUP 4 (ultra-processed) and any product that contains an isolated protein (in any amount) will also be categorised as ultra-processed.
Example 2: Granulated cane sugar is a highly processed food. Indeed, the processes it undergoes are similar to pharmaceutical-grade products. But under NOVA, refined sugar is not ultra-processed and is not considered to be a substance that characterises ultra-processed foods. On the other hand Stevia (a sugar substitute extracted from the plant Stevia rebaudiana) is classified as GROUP 4 (ultra-processed) and any food that contains this product (in any amount) will also belong to GROUP 4.
Example 3: White refined flour and instant coffee are categorised as GROUP 1 (minimally processed): this is puzzling.
Example 4: A cake made from white (refined) flour, granulated sugar, isolated starches, refined vegetable oil, baking soda (E500) and baking powder (E331) will not be ultra-processed if it does not contain any "unusual" ingredients or "cosmetic additives". But a bread made from wholemeal unrefined flour, yeast, water and soy lecithin (E322) is de facto ultra-processed (on account of the soy lecithin, which is considered a "cosmetic additive").
NOVA's system of rules is illogical. The problem is that NOVA classifies "traditional" products (e.g, isolated sugars) differently to "unusual" products (e.g., isolated protein) even when their level of processing is comparable. I speculate that this is related to an intuition that "traditional" ingredients are less problematic compared to those developed more recently. But why should we given any credence to such appeals to tradition? I submit that we shouldn't.
Here is a problem which, I think, will apply to any system that categorises foods on the basis of their level of processing:
Example 5: Home-made dishes need to be broken down into their ingredients before being assigned to a NOVA category. This entails that reproducing an ultra-processed food at home (using the exact same ingredients) will not contribute to the amount of ultra-processed foods you eat (because only the "cosmetic additives" and "untraditional ingredients" will be classified as ultra-processed).
3. The Associations Between UPF and adverse health outcomes are driven by a handful of foods.
Numerous prospective cohort studies have linked higher consumption of UPFs to an increased risk of various diseases. Moreover, these associations tend to persist after adjustment for diet quality3. Consequently, many nutrition scientists are concerned that UPFs may be problematic independently of established nutritional factors (e.g., saturated fat, fibre, added sugar, energy density, palatability).
However, studies that investigate the relationship between subclasses of UPF and bad health outcomes typically find that only a handful of foods are associated with statistically significant increases in risk. The foods mostly commonly associated with increased disease incidence appear to be ultra-processed drinks and ultra-processed animal products4,5,6,7,8,9,10,11,17,18 – foods already known (or strongly suspected) to be bad for human health. Most other UPFs are either not associated or inversely associated with bad health outcomes.
Thus, it seems to me irrational to fear (or even to be wary of) all ultra-processed foods on the basis of the observational research. Food and drinks should be judged on a case-by-case basis, using the highest quality evidence available. In my opinion, the fact that a food happens to be categorisable as ultra-processed should be considered irrelevant when evaluating its impact on human health.
4. Hall et al. 2019 is Uncompelling
In 2019 Kevin Hall and colleagues12 published a randomised controlled trial on ultra-processed foods. Very briefly, this study found that under ad libitum conditions participants consumed more calories and gained more weight when presented with ultra-processed foods (compared to minimally processed foods) for 14 days. A key strength of this study is that it's a crossover RCT metabolic ward trial (considered by many to be "gold standard" in nutrition). Moreover, the two diets were matched for "calories, sugar, fat, sodium, fiber and macronutrients".
So does this study advance strong evidence in favour of the idea that ultra-processed foods are inherently obesogenic? In my opinion, absolutely not.
The problem is that the meals given during the ultra-processed and minimally processed conditions were totally different. In order to isolate the impact of ultra-processing, everything should be held constant (food type, ingredients, flavour, texture, calories, macros etc.) except processing. Take a look at the examples of the kind of meals served in this study:
It's hardly surprising that the meal on the left was less sating. The ultra-processed foods were higher in energy density, and lower in fibre. In order equalise the meals in these respects, participants were given several glasses of artificially sweetened and fibre-enriched drinks. But as the authors note, beverages appear to have a very limited capacity to impact satiety. Other differences includes lower food volume and more added sugar under the ultra-processed condition.
5. "Ultra-Processed People" is Uncompelling
This book13 by Chris van Tullekan has become a best-seller and captured the imagination of the public. Many appear to be convinced that UPFs are intrinsically hazardous to health.
But like so many popular nutrition science books, the overall message is not credible.
Problem 1: Dr van Teullekan describes a self-experiment whereby he significantly increases the proportion of ultra-processed foods in his diet. Firstly, N=1 experiments are extremely low-quality evidence. Secondly, he based his diet mostly on junk foods and therefore we would expect his health markers to worsen due known problems with these foods. It tells us nothing about the impact of ultra-processing as such.
Problem 2: He relies heavily on observational studies. As explained above, the associations appear to be driven by a limited number of foods already known (or strongly suspected) to be problematic.
Problem 3: He presented quotes from authors and activists that are convinced that the evidence on UPF is sufficient to warrant immediate action and rarely mentioned more critical viewpoints. This may mislead readers to think that there is a scientific consensus on the harms of UPFs. But in fact, numerous credible organisations hold that for the time being dietary recommendations should continue to rely on traditional compositional factors (examples: The British Nutrition Foundation14, Scientific Advisory Committee on Nutrition15, The British Dietetic Association16 among many others).
Laying my Cards on the Table
Several key healthy animal-free foods are judged to be ultra-processed and this is why (as a vegan and advocate for animal-free diets) I am so motivated to combat anti-UPF viewpoints. Now to be clear, I have no problem accepting that there are many vegan junk foods on the market. But I firmly reject the idea that items such as commercially available soy milk, packaged wholegrain breads and well-formulated animal-free meats (and so on) should be demonised or treated with caution merely because:
1) They happen to be placed in a category alongside junk foods.
2) They are untraditional.
3) Epidemiologists have shown that ultra-processed drinks and ultra-processed animal products are reliably associated with harm.
4) A study found that energy dense, HFSS foods were not as sating as meals based on whole foods.
5) A popular book created a damning narrative around UPFs.
References
1. Monteiro, C.A., Cannon, G., Levy, R.B., Moubarac, J.C., Louzada, M.L., Rauber, F., Khandpur, N., Cediel, G., Neri, D., Martinez-Steele, E. and Baraldi, L.G., 2019. Ultra-processed foods: what they are and how to identify them. Public health nutrition, 22(5), pp.936-941.
2. https://npmcalculator.cdrc.ac.uk/
3. Dicken, S.J. and Batterham, R.L., 2021. The role of diet quality in mediating the association between ultra-processed food intake, obesity and health-related outcomes: a review of prospective cohort studies. Nutrients, 14(1), p.23.
4. Chen, Z., Khandpur, N., Desjardins, C., Wang, L., Monteiro, C.A., Rossato, S.L., Fung, T.T., Manson, J.E., Willett, W.C., Rimm, E.B. and Hu, F.B., 2023. Ultra-processed food consumption and risk of type 2 diabetes: three large prospective US cohort studies. Diabetes Care, 46(7), pp.1335-1344.
5. Wang, L., Pan, X.F., Munro, H.M., Shrubsole, M.J. and Yu, D., 2023. Consumption of ultra-processed foods and all-cause and cause-specific mortality in the Southern Community Cohort Study. Clinical Nutrition, 42(10), pp.1866-1874.
6. Juul, F., Vaidean, G., Lin, Y., Deierlein, A.L. and Parekh, N., 2021. Ultra-processed foods and incident cardiovascular disease in the Framingham Offspring Study. Journal of the American College of Cardiology, 77(12), pp.1520-1531.
7. Hang, D., Wang, L., Fang, Z., Du, M., Wang, K., He, X., Khandpur, N., Rossato, S.L., Wu, K., Hu, Z. and Shen, H., 2023. Ultra-processed food consumption and risk of colorectal cancer precursors: results from 3 prospective cohorts. JNCI: Journal of the National Cancer Institute, 115(2), pp.155-164.
8. Du, S., Kim, H., Crews, D.C., White, K. and Rebholz, C.M., 2022. Association between ultraprocessed food consumption and risk of incident CKD: a prospective cohort study. American Journal of Kidney Diseases, 80(5), pp.589-598.
9. Li, H., Li, S., Yang, H., Zhang, Y., Zhang, S., Ma, Y., Hou, Y., Zhang, X., Niu, K., Borné, Y. and Wang, Y., 2022. Association of ultraprocessed food consumption with risk of dementia: a prospective cohort study. Neurology, 99(10), pp.e1056-e1066.
10. Osté, M.C., Duan, M.J., Gomes-Neto, A.W., Vinke, P.C., Carrero, J.J., Avesani, C., Cai, Q., Dekker, L.H., Navis, G.J., Bakker, S.J. and Corpeleijn, E., 2022. Ultra-processed foods and risk of all-cause mortality in renal transplant recipients. The American Journal of Clinical Nutrition, 115(6), pp.1646-1657.
11. Zhao, L., Clay-Gilmour, A., Zhang, J., Zhang, X. and Steck, S.E., 2023. Higher ultra-processed food intake is associated with adverse liver outcomes: a prospective cohort study of UK Biobank participants. The American Journal of Clinical Nutrition.
12. Hall, K.D., Ayuketah, A., Brychta, R., Cai, H., Cassimatis, T., Chen, K.Y., Chung, S.T., Costa, E., Courville, A., Darcey, V. and Fletcher, L.A., 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), pp.67-77.
13. van Tulleken, C., 2023. Ultra-Processed People: Why Do We All Eat Stuff That Isn't Food... and Why Can't We Stop?. Random House.
14 .Lockyer, S., Spiro, A., Berry, S., He, J., Loth, S., Martinez?Inchausti, A., Mellor, D., Raats, M., Sokolovi?, M., Vijaykumar, S. and Stanner, S., 2023. How do we differentiate not demonise–Is there a role for healthier processed foods in an age of food insecurity? Proceedings of a roundtable event. Nutrition bulletin.
15. https://www.gov.uk/government/publications/sacn-statement-on-processed-foods-and-health/sacn-statement-on-processed-foods-and-health-summary-report
16. https://www.bda.uk.com/resource/processed-food.html
17. Li, H., Li, S., Yang, H., Zhang, Y., Ma, Y., Hou, Y., Zhang, X., Sun, L., Borné, Y. and Wang, Y., 2023. Association of Ultra?Processed Food Intake with Cardiovascular and Respiratory Disease Multimorbidity: A Prospective Cohort Study. Molecular Nutrition & Food Research, 67(11), p.2200628.
18. Canhada, S.L., Vigo, Á., Levy, R., Luft, V.C., da Fonseca, M.D.J.M., Giatti, L., Molina, M.D.C.B., Duncan, B.B. and Schmidt, M.I., 2023. Association between ultra-processed food consumption and the incidence of type 2 diabetes: the ELSA-Brasil cohort. Diabetology & Metabolic Syndrome, 15(1), pp.1-10.
