New evidence connects ultra-processed foods to rising diabetes rates

By Dr. Priyom Bose, Ph.D.Reviewed by Benedette Cuffari, M.Sc.Sep 18 2024

Unraveling the impact of food processing on diabetes: can swapping ultra-processed foods with healthier alternatives reduce type 2 diabetes risk?

Study: Food consumption by degree of food processing and risk of type 2 diabetes mellitus: a prospective cohort analysis of the European Prospective Investigation into Cancer and Nutrition (EPIC). Image Credit: Discover 4K World / Shutterstock.com

Extensive research has established that consuming ultra-processed food (UPFs) negatively impacts the development of type 2 diabetes mellitus (T2D); however, it remains unclear how other degrees of food processing affect T2D risk. A recent Lancet Regional Health study examines this association.

Food processing and T2D risk

The Nova classification system categorizes food as unprocessed/minimally processed foods (MPFs), processed culinary ingredients (PCIs), processed foods (PFs), and UPFs. Higher UPF consumption has been associated with weight gain, obesity, T2D, and cardiovascular disease. Likewise, one study utilizing the European Prospective Investigation into Cancer and Nutrition (EPIC) reported the association between UPF intake and risk of cancer and cardiometabolic multimorbidity.

However, the association between T2D risk and all Nova groups remains unclear. Thus, future studies are needed to determine whether UPFs or other food products that have undergone different degrees of processing are associated with the least favorable outcomes. Clarifying the differential effects of these processed foods could allow researchers to quantify the benefits of replacing UPFs with MPFs, PCIs, or PFs.

About the study

The current prospective cohort study involved an analysis of EPIC data, in which dietary intake was assessed at baseline by utilizing dietary questionnaires. Food consumption patterns were subsequently categorized according to the Nova food classification system. Multiple sources were used to verify T2D cases including self-report, primary- and secondary-care registers, hospital admissions, medications registers, and mortality data.

To estimate the associations between T2D and MPFs + PCIs, PFs and UPF intake, Cox regression and statistical substitution analyses were perfromed. UPF sub-group analyses were also conducted to determine heterogeneity in the association between T2D and UPFs. In each analysis, different reference or control groups were used.

The mean age of the study participants was 52.5 years, 63.5% of whom were female and the average body mass index (BMI) was 25.7 kg/m². During the mean follow-up period of 10.9 years, 4.6% of the participants received a T2D diagnosis. The average percentage contribution of MPF + PCIs, UPFs, and PFs to total daily dietary intake in grams was 72.1%, 13.0%, and 14.9%, respectively.

Study findings

Concerning the relationships between Nova group intakes and incident T2D,

Restricted cubic spline analyses identified linearity across MPFs + PCIs until high intakes. In a second adjusted model, higher consumption of MPFs + PCIs and PFs was associated with lower incident type 2 diabetes mellitus, whereas a higher risk of T2D was associated with higher UPF intake.

Statistical substitution analyses were performed to elucidate the impact of replacing 10% g/day from UPFs, PFs, or MPFs + PCIs for another food group on incident T2D. Replacing 10% g/day of UPFs with MPFs + PCIs or PFs, as well as replacing MPFs + PCIs with PFs, was found to reduce the incidence of T2D in adjusted models. These substitution results were similar after adjusting for covariate factors such as saturated fat, sugar, and Mediterranean diet adherence.

The association between each UPF sub-group and incident T2D was investigated to explore heterogeneity in the association between T2D and UPF. Each 10% g/day higher intake of savory snacks, ready-to-eat/heat mixed dishes, animal-based products, and artificially- and sugar-sweetened beverages (ASB/SSB) was associated with higher incident T2D.

Lower incident T2D was observed when considering the consumption of sweets and desserts, biscuits and breakfast cereals, plant-based alternatives, and breads. No association between T2D and sauces, spreads, condiments, alcoholic beverages, and other UPFs was identified.

Mediation analysis was performed on the waist-height ratio (WHtR), an adiposity measurement, to explore the potential factors that could explain the association between UPF and incident T2D. To this end, WHtR was found to mediate 46.4% of the association between T2D and UPF.

These findings remained consistent in the sensitivity analyses; however, a statistically significant association was not observed between T2D and UPF consumption in France or Italy in these analyses. Furthermore, MPFs + PCIs was not associated with lower incident T2D when the intake was modeled as kcal/day, %kcal/day, or g/day.

Conclusions

Higher consumption of UPFs was associated with higher incident T2D, whereas the increased consumption of less processed foods was associated with lower incident T2D.

Significant heterogeneity was observed within UPFs; however, measurement errors or unmeasured confounding factors may have contributed to this observation.

Taken together, these findings emphasize the importance of reducing the consumption of specific UPFs, rather than using an overall UPF metric for public guidance.