Summary
In a cohort of 461,682 older US adults, total ultra-processed food intake was not associated with colorectal cancer incidence. Associations differed for specific dietary components, including meat, whole grains, fibre and dairy.
A prospective study of 461,682 US adults found no association between total ultra-processed food (UPF) intake and colorectal cancer incidence over a median 18.8 years of follow-up. The analysis, published in PLOS Medicine, also found that associations varied by the foods and nutrients counted within the broad UPF category: meat intake was associated with higher incidence, while whole grains, fibre, dairy and calcium were associated with lower incidence.
The findings come from the NIH-AARP Diet and Health Study and concern colorectal cancer in older US adults—not cancer risk in general. They highlight why a single processing category can combine foods with different nutritional characteristics.
What the cohort found
Participants were aged 50–71 when they joined the study in 1995–96. Researchers analysed those who were cancer-free at baseline and followed cancer incidence through 2018. During follow-up, 10,075 colorectal adenocarcinomas were diagnosed. Participants reported their diets using a food-frequency questionnaire, which researchers used to estimate intake and classify foods under the Nova system. Nova group 4 comprises ultra-processed foods.
Comparing the highest and lowest fifths of energy-adjusted UPF intake, the estimated hazard ratio for colorectal cancer was 0.97 (95% confidence interval 0.91–1.03; trend test p=0.55). In other words, the study found no overall association between total UPF intake and incidence. Results were also broadly similar across cancer locations in the colon and rectum.
The researchers then examined specific food groups and nutrients from UPF and non-UPF sources. In models adjusting for both sources, higher meat intake was associated with higher incidence whether it came from UPF sources (hazard ratio 1.08 per serving per day; 95% CI 1.02–1.14) or non-UPF sources (1.04; 1.01–1.06). Dairy and calcium intake from both types of source were associated with lower incidence. Whole-grain intake from UPF sources was associated with lower incidence, as was fibre intake from UPF sources; the corresponding non-UPF estimates were not statistically significant in those models.
A separate model estimated that replacing 10% of food weight from UPF with 10% from minimally processed foods was associated with a 2% lower colorectal cancer incidence (hazard ratio 0.98; 95% CI 0.96–0.99). This was a statistical substitution analysis of reported diets, not a trial in which participants changed what they ate.
Why food composition matters
UPF is a broad processing-based category, not a uniform measure of nutritional content. In this cohort, products classified as UPF contributed to intake of both dietary components associated with lower colorectal cancer incidence—such as whole grains and dairy—and meat, which was associated with higher incidence. The researchers suggest that this mix may help explain why studies of total UPF intake and colorectal cancer have produced differing results.
The study supports a distinction between the degree of food processing and the composition of a diet. Its results are consistent with dietary guidance cited by the authors to consume whole grains, fibre, dairy and calcium and to avoid processed meat for colorectal cancer prevention. Because this was an observational cohort study, the associations do not show that these foods or nutrients caused the differences in cancer incidence.
The evidence also has limits relevant to interpretation. Participants reported diet at baseline, and the food questionnaire was not designed to measure processing levels directly; researchers linked reported foods to population food databases to classify them. The cohort consisted of older, mostly white US adults recruited in the mid-1990s, so the findings may not apply in the same way to younger or more diverse populations, or to current diets.