Some ultra-processed foods may reduce risk of developing diabetes

From both a research and policy perspective, interest in understanding the effect food processing has on health outcomes continues to grow. To this point, the scientific advisory committee for the 2025-2030 US Dietary Guidelines has been asked to address this question “What is the relationship between consumption of dietary patterns with varying amounts of ultra-processed foods [UPFs] and growth, body composition, and risk of obesity?” Also, the USDA recently held a two-day workshop focused on developing research recommendations aimed at better understanding the impact of UPFs on obesity and cardiovascular disease. The term UPF was coined in 2009 when Brazilian researchers created the Nova food classification. Nova classifies all foods into one of 4 groups based entirely on the extent to which they have been processed.

For the past several years, Soy Nutrition Institute Global has carefully monitored the discussion and debate around UPFs. We have written about and given presentations on Nova and UPFs because these topics are of considerable interest to the soy food industry. Soy-based meat alternatives and most soymilks (including those based on whole soybeans) are classified by Nova as UPFs, foods which many increasingly argue should be limited in the diet if not completely avoided. This is why new research by Chen and colleagues published in Diabetes Care showing that the intake of some UPFs is associated with a reduced risk of type 2 diabetes (T2D) is so relevant.

Chen and colleagues examined the relationship between UPF intake and risk of T2D among participants in one of three US cohorts: the Nurses’ Health Study (n=71,871), the Nurses’ Health Study II (n=87,918) and the Health Professional Follow-Up study (n=38,847). Over the several decade-long follow up period, 19,503 participants developed T2D. After adjustment, the pooled hazard ratios (versus quintile 1, reference) for UPF intake quintiles 2-5 were 1.09, 1.19, 1.20 and 1.28, respectively (p for trend, <0.0001). It is notable that in each of the 3 cohorts, there was a statistically significant dose response relationship between UPF intake and T2D risk. That is, as UPF intake increased, so did risk.

However, sub-analysis revealed striking differences in associations among the intake of various types of UPFs. For example, the hazard ratio (95% CI) for animal-based UPFs was 1.44 (1.38, 1.51) whereas for fruit-based UPFs it was 0.82 (0.77, 0.86). For ready-to-eat/heat mixed dishes and packaged savory snacks, these values were 1.42 (1.31, 1.54) and 0.91 (0.87, 0.94), respectively.

These findings highlight the importance of not painting all Nova-classified UPFs with a broad brush. Further, they show some UPFs are not only not harmful, but potentially beneficial.

This finding may not be surprising to those familiar with a 2022 paper entitled “Perspective: Soy-based meat and dairy alternatives, despite classification as ultra-processed foods, deliver high-quality nutrition on par with unprocessed or minimally processed animal-based counterparts.” I along with the other co-authors concluded that none of the common concerns about UPFs apply to soy burgers and soymilk more than they do to their animal-based counterparts, meat and milk, which are Nova-classified as unprocessed or minimally processed foods — foods to be encouraged in the diet. These concerns dealt with palatability, satiety, caloric density and eating rate, among several others.

We went on to conclude that “… in the case of soy-based meat alternatives and soymilks, the NOVA classification system is overly simplistic and of little utility for evaluating the true nutritional attributes of these foods.” This conclusion is consistent with sentiment expressed by Vadiveloo and Gardner, in a commentary published in response to the research by Chen et al., when they stated that not all ultra-processed foods are created equal.

The lack of nuance inherent to the Nova food classification system greatly limits its utility. And any food classification system that doesn’t differentiate soy burgers and soymilk from twinkies, does a disservice to the public.

 This blog is sponsored by SNI Global and U.S. Soy.