Summary: In a cross-sectional analysis in type 2 diabetes, higher processed and ultra-processed food intake was not significantly linked to glucose or body-size measures, but intermediate and high consumption was associated with higher total and LDL cholesterol; the design cannot establish causation.
PICO Summary
| Element | Detail |
|---|---|
| Population | 326 patients with type 2 diabetes; baseline cross-sectional data from the multicentre NUGLIC trial, Brazil. |
| Intervention | Higher intake of processed and ultra-processed foods (NOVA classification; UPF was ~16.4% of daily energy). |
| Comparison | Lower intake (consumption quintiles), against HbA1c as the primary outcome. |
| Outcome | No significant linear association with HbA1c, fasting glucose, BMI, or waist circumference. Intermediate and high consumption was associated with higher total cholesterol (Q3 β=26.6; Q4 β=26.7) and LDL cholesterol (Q4 β=19.8; Q5 β=17.5). Cross-sectional design precludes causal inference. |
Expert Commentary
This is a useful reality check whose findings should be reported precisely rather than folded into a general anti-processed-food message. Despite the strong prior that ultra-processed foods worsen metabolic health, this analysis found no significant association between their consumption and the glycaemic or anthropometric markers that define diabetes control, no clear link with HbA1c, fasting glucose, BMI, or waist. What did emerge was an association with higher total and LDL cholesterol at the upper consumption levels, which is biologically plausible and clinically relevant given cardiovascular risk in diabetes, but it is one signal among several null ones. The design is the decisive limitation: a cross-sectional snapshot cannot tell us whether ultra-processed intake raised cholesterol or whether other confounders drive both, and the authors are appropriately careful to disclaim causality. Can I use this with my patients? Modestly and honestly. It supports advising diabetic patients to limit ultra-processed foods primarily for their lipid and cardiovascular implications, while I avoid overstating the case, this particular dataset did not show a glycaemic penalty, and only a prospective or interventional study could establish cause and effect.
References
Bauer J, Ayala FO, Marcadenti A, et al. Consumption of ultra-processed foods and metabolic parameters in type 2 diabetes mellitus: a cross-sectional study. Int J Environ Res Public Health. 2025;22(8):1275. doi:10.3390/ijerph22081275