Study: ultra-processed foods raise diabetes risk

A 2025 meta-analysis is the source behind social posts claiming that ultra-processed foods are linked to a higher risk of type 2 diabetes. The paper, published online on June 9, 2025 in *Diabetes & Metabolism Journal*, pooled 12 prospective cohort studies and found that people in the highest intake categories had a summary relative risk of 1.48 versus those in the lowest categories. The same paper reported a 1.14 relative risk for every 10% increase in ultra-processed food intake measured by grams per day, and said the dose-response curve became steeper above about 300 grams a day. ### Where did the “48% higher risk” claim come from? The “48% higher risk” figure comes from the paper’s comparison of the highest versus lowest intake categories. In the abstract, the authors reported a summary relative risk of 1.48, with a 95% confidence interval of 1.36 to 1.61, for type 2 diabetes among people with the highest ultra-processed food intake compared with those with the lowest intake. (e-dmj.org) A relative risk of 1.48 is commonly described as a 48% higher risk, but it is not the same thing as saying 48% of high consumers will develop diabetes. The paper was summarizing differences in risk between groups across cohort studies, not giving an individual prediction. That is an inference from the reported statistic. ### What exactly did the study measure? (e-dmj.org) The authors said they searched PubMed, Embase and Web of Science for prospective studies published through January 2024. After screening 569 publications, they included 12 prospective cohort studies in the dose-response meta-analysis. The paper examined ultra-processed food intake in several ways: percentage of grams per day, absolute grams per day, and servings per day. (e-dmj.org) For each 10% increase in intake measured as grams per day, the summary relative risk was 1.14. For each additional 100 grams per day, the summary relative risk was 1.05, and for each additional serving per day it was 1.04. ### What does the “300 grams a day” line mean? The paper said its nonlinear dose-response analysis suggested a steeper increase in diabetes risk at more than about 300 grams per day of ultra-processed foods. That is the source for posts saying the effect intensified past 300 grams daily. The study did not present that threshold as a hard cutoff for any one person. (e-dmj.org) It described the shape of the pooled association across studies, based on restricted cubic spline analysis. ### What counts as an ultra-processed food here? The underlying cohort studies classified foods using the NOVA system, which groups foods by the nature and extent of processing. (e-dmj.org) NIH has described ultra-processed foods under NOVA as industrial formulations, and Harvard’s Nutrition Source says the system distinguishes between degrees of processing rather than treating all processed foods the same way. The 2025 meta-analysis also reported that the association was not uniform across all subgroups. Processed meats were positively associated with diabetes risk, while ultra-processed cereals and breads and packaged savory snacks showed inverse associations in the pooled subgroup analysis. ### Is this the only study on the topic? A 2023 *Diabetes Care* paper using three large U.S. cohorts reported a hazard ratio of 1.46 when comparing extreme quintiles of total ultra-processed food intake, based on 19,503 type 2 diabetes cases over 5,187,678 person-years. (nih.gov) A 2022 systematic review and meta-analysis in the *International Journal of Epidemiology* also found higher risk with greater intake, reporting a 12% higher risk for moderate intake and 31% higher risk for high intake versus non-consumption. (e-dmj.org) The June 9, 2025 paper remains the apparent source for the specific figures circulating in social posts — 48% for highest versus lowest intake, 14% per extra 10% grams per day, and a steeper curve above roughly 300 grams daily. (pubmed.ncbi.nlm.nih.gov) (e-dmj.org) (pubmed.ncbi.nlm.nih.gov)