Ultra-Processed Food Showed No Overall Link to Colorectal Cancer, but Its Parts Pulled Both Ways

Researchers at the National Cancer Institute reported Oct. 15, 2026, in PLOS Medicine that older US adults who ate the most ultra-processed food were no more likely to develop colorectal cancer than those who ate the least. The analysis covered 461,682 people who were free of cancer when they joined the NIH-AARP Diet and Health Study, 10,075 of whom were later diagnosed with the disease.

Earlier studies of the question have disagreed. Leila Abar and Caitlin P. O'Connell of the National Cancer Institute, with senior author Erikka Loftfield, wrote that the mix of foods inside the ultra-processed category may be the reason: it holds foods tied to lower colorectal cancer risk, such as whole grain products and dairy, next to foods tied to higher risk, such as processed meat. Their study measured the category as a whole and those components separately.

The authors reported a hazard ratio of 0.97 for the fifth of participants eating the most ultra-processed food against the fifth eating the least, where 1.0 would mean no difference between the groups. The 95% confidence interval ran from 0.91 to 1.03, a range covering a small reduction in risk and a small increase alike, so the cohort rules out a large effect in either direction rather than showing there is none. Diet was recorded once, on a questionnaire at enrollment, and cases were counted through 2018, a median of 18.8 years later.

Individually, the components pulled opposite ways. The team found that whole grain, fiber, dairy and calcium from ultra-processed sources went with lower colorectal cancer incidence, while meat from those sources went with higher. These are associations drawn from diets people reported themselves, and the authors say the component results should be read cautiously because many comparisons were made. Their conclusion is that the findings support existing advice to eat whole grains, fiber, dairy and calcium and to avoid processed meat.

Funding for the work came from the National Cancer Institute's intramural research program. The analysis code is posted publicly on GitHub, and the cohort data are available to researchers who apply to the study's steering committee.