Avoid ultra-processed foods! They are worse than run-of-the mill processed products, we read. The food media embraced the concept that ultra-processed foods cause chronic disease. Many studies suggested this conclusion. Food journalists jumped on the bandwagon. Is this the magic bullet that will lead to a healthy, more fit America? Will it lead to a healthier, more fit world? It must be so. I read it online.
A few food scientists urged caution, but their message failed to reach a general audience. Michael Gibney noted that the classification of ultra-processed foods varies (1). Coding in some studies for foods in the category do not match with items in other studies. Lack of consistency obscures the meaning of such investigations. Julie Jones and Roger Clemens point out that the class of ultra-processed food is confusing (2). It is difficult to separate ultra-processing from normal processing. Correlation not experimentation characterize almost all negative studies on the topic. The implication is that cause and effect between UPFs and chronic disease is clear. Correlation is not causation. Early research associated obesity (3), cancer (4), and heart disease (5) with UPFs. One study even suggests that eating UPFs relates to all-cause mortality (6). What is that supposed to mean? Is our diet the only important lifestyle choice in the matter of life and death?
More recent articles in scientific journals push back on NOVA. They are not convinced that UPFs cause chronic disease. The relationships are not as clear as it may appear. There are at least six areas of concern that suggests the need for further discussion.
Consumers do not understand what foods qualify as ultra-processed (7). Journalists have found a powerful phrase that scares consumers. Does it matter that the term also confuses their audience? Most online articles point out ultra-processed foods with fat, salt, and sugar as the villains. Homemade foods high in fat, salt, and sugar are OK. Only culinary ingredients are familiar and permitted in the home kitchen. Those additives used in industrial formulations “are not usually available in supermarkets (8).” Note that many of these prohibited ingredients are available online.
Food professionals do not understand what an ultra-processed food is. Food professionals could not differentiate processed from ultra-processed foods (9). Another study involved online discussions among professionals across many disciplines (10). Basic concepts such as what does it mean to “process” foods were in contention. Without a consensus on key concepts, how can we expect consumers to understand?
The ultra-processed category is too broad. Can we divide all foods into four categories and condemn one of those categories? Group 4, ultra-processed foods, provides 20% of Italy’s calories and over 60% of those in the United States. Does the classification scheme help us separate out true causes of chronic disease? An article relating UPFs to diabetes found the category too broad (11). It concluded that there was a relationship between UPFs and the disease. They noted that the breadth of the category obscured the effect.
Ultra-processed foods have no subcategories. If they did, we could separate potential problem foods from ones that are not. A systematic review found some subcategories were more problematic than others (12). Breakfast cereals, for example, were beneficial to health rather than detrimental. They tagged sweetened beverages, either by sugar or artificial sweeteners as problematic. They also noted that cured meat and red meat contributed to all-cause mortality. Other lifestyle factors were co-variates making conclusions difficult. Further studies must adopt standard methods for better data analysis.
The reliability of the data is suspect. Are studies on ultra-processed foods reliable? NOVA dropped the concept of more than 5 ingredients to characterize ultra-processed foods. It is not clear that the target foods changed with this modification. Plant-based meats and milks fit the definition. Many of these products came into being after the initial guidelines appeared. It is not clear if researchers coded for these types of products. Another systematic review on the topic (13) questioned the reliability of survey data.
The threat on the horizon is development of the next Dietary Guidelines. Efforts to consider ultra-processed products in developing the guidelines are underway. Will the committee consider only negative studies based on correlation on large databases? Or will it balance previous work with more recent studies highlighted in this post? We must be cautious in drawing conclusions from correlation with large databases (14). In the past, the committee has been slow to make drastic changes in its recommendations. A careful assessment of the long-range implications of major changes is necessary. The joke is that few consumers use the guidelines to design personal diets. Food journalists will be looking at the decisions. Most will either praise the new guidelines for the courage to condemn UPFs. Or, they will criticize the guidelines for failure to condemn them.
Bottom line. So, what do we have here? Ultra-processed is a term that neither consumers nor most food professionals understand. Neither group can place processed foods into the proper categories. It is a useful term to market unsavory thoughts stigmatizing many processed foods. NOVA casts a wide net, grouping many foods that are not related to each other as ultra-processed. Correlation of massive databases obscures which foods might be dangerous to health. Is every ultra-processed food causing obesity, diabetes, and heart disease? Shouldn’t we break the category down into subgroups before condemning all UPFs?
The discussion is shifting. We need to take a broader view on the data collected to date. Lets not make dramatic changes in the Dietary Guidelines until we have a clearer picture. Earlier momentum should not rule discussions without an appreciation of more recent information. Read Christine Rosenbloom’s thoughtful, nuanced discussion on ultra-processed foods
* Probable NOVA classification of foods in the photo: Group One (Unprocessed or minimally processed): baby carrots, wheat berries; Group Three (Processed): peanut butter, canned spinach, beer, pasta; Group Four (Ultra-processed): bacon, peanut butter cups, kimchi, soda, margarine, hummus; Uncertain (depending on additives): California rolls
Next week: Ending Hunger in the Age of Climate Change
(1) Gibney, M.J. 2019. Ultra-Processed Foods: Definitions and Policy Issues Current Developments in Nutrition, Volume 3, Issue 2, nzy077, https://doi.org/10.1093/cdn/nzy077
(2) Jones. J.M. and R.A. Clemens, 2017. Introductory brain teaser for the cereal chemist—How do we categorize processed and ultra-processed foods? Cereal Foods World 63 (4): 182-184.
(3) Nardocci, M., J.Y. Polsky, and J. C. Moubarac, JC. 2021. Consumption of ultra-processed foods is associated with obesity, diabetes and hypertension in Canadian adults. Can J Public Health 112, 421–429. https://doi.org/10.17269/s41997-020-00429-9
(4) Fiolet, T., B. Srour, L. Sellem, E. Kesse-Guyot, B. Allès, C. Méjean, M. Deschasaux, P. Fassier, P. Latino-Martel, M. Beslay, S. Hercberg, C. Lavalette, C.A. Monteiro, C. Julia, and M. Touvier, 2018. Consumption of ultra-processed foods and cancer risk: results from NutriNet-Santé prospective cohort. BMJ 2018; 360 https://doi.org/10.1136/bmj.k322
(5) Zhong, GC., H.T. Gu, Y. Peng, K. Wang, Y-Q-L. Wu, T-Y Hu, F-C. Jing, and F.B. Hao, 2021. Association of ultra-processed food consumption with cardiovascular mortality in the US population: long-term results from a large prospective multicenter study. Int J Behav Nutr Phys Act 18, 21 (2021). https://doi.org/10.1186/s12966-021-01081-3
(6) A. Rico-Campà, M.A. Martínez-González, I. Alvarez-Alvarez, R. de Deus Mendonça, C. de la Fuente-Arrillaga, C. Gómez-Donoso, M. Bes-Rastrollo, 2019. Association between consumption of ultra-processed foods and all cause mortality: SUN prospective cohort study. BMJ, https://doi.org/10.1136/bmj.l1949
(7) Sarmiento-Santos, J., M.B.N. Souza, L.S. Araujo, J.M. Pion, R.A. Carvalho, and F.M. Vanin. 2002. Consumers’ Understanding of Ultra-Processed Foods. Foods 2022, 11(9), 1359; https://doi.org/10.3390/foods11091359
(8) Nestle, M. 2022. Regulating the food industry: An aspirational agenda. American Journal of Public Health 112:853-858 https://ajph.aphapublications.org/doi/10.2105/AJPH.2022.306844
(9) Braesco, V., I. Souchon, P. Sauvant, T. Haurogné, M. Maillot, C. Féart, and N Darmon. 2022, Ultra-processed foods: how functional is the NOVA system?. Eur J Clin Nutr . https://doi.org/10.1038/s41430-022-01099-1
(10) Sadler, C.R., T. Grassby, K. Hart, M.M. Raats, M. Sokolović, and L. Timotijevic, 2022. “Even We Are Confused”: A Thematic Analysis of Professionals’ Perceptions of Processed Foods and Challenges for Communication. Front Nutr. 2022; 9: 826162. doi: 10.3389/fnut.2022.826162
(11) Duan, M-J, P.C. Vinke, G. Navis, E. Corpeleijn and L.H. Dekker, 2022. Ultra-processed food and incident type 2 diabetes: studying the underlying consumption patterns to unravel the health effects of this heterogeneous food category in the prospective Lifelines cohort. BMC Med 20, 7 (2022). https://doi.org/10.1186/s12916-021-02200-4
(12) Taneri, P.E., F. Wehrli, Z.M. Roa-Díaz, O.A. Itodo, D. Salvador, H. Raeisi-Dehkordi, L. Bally, B. Minder, J.C. Kiefte-de Jong, J.E Laine, A. Bano, M. Glisic, and T. Muka, 2022. Association Between Ultra-Processed Food Intake and All-Cause Mortality: A Systematic Review and Meta-Analysis. Amer. J. Epidemiol., kwac039, https://doi.org/10.1093/aje/kwac039
(13) Pagliai, G., M. Dinu, M.P. Madarena, M. Bonaccio, L. Iacoviello and F. Sofi, 2020. Consumption of ultra-processed foods and health status: a systematic review and meta-analysis. British Journal of Nutrition, 125(3): 308 – 318. https://doi.org/10.1017/S0007114520002688
(14) Austin, P.C., Mamdani, M.M., Juurlink, D.N. and Hux, J.E. 2006. Testing multiple statistical hypotheses resulted in spurious associations: a study of astrological signs and health. J. Clin. Epidemiol. 59:964-969. https://doi.org/10.1016/j.jclinepi.2006.01.012