Ultra-Processed Foods Show No Clear Link to Colorectal Cancer in Massive US Cohort Study

A sweeping analysis of dietary data from nearly half a million older American adults has found that total intake of ultra-processed foods was not associated with the risk of developing colorectal cancer, challenging a simple narrative in which these industrially formulated products are uniformly blamed for one of the most common malignancies in the Western world. The study, published in PLOS Medicine, drew on participants in the NIH-AARP Diet and Health Study and followed them for more than two decades, during which more than ten thousand cases of colorectal adenocarcinoma were diagnosed. Its central finding is nuanced: while overall ultra-processed food consumption showed no measurable link to cancer incidence, individual components delivered within these foods pointed in opposite directions, with some raising risk and others lowering it.

Ultra-processed foods, classified under the widely used Nova system, account for more than half of the daily energy intake of adults in the United States. These products, which include packaged snacks, ready-to-eat meals, soft drinks, and reconstituted meats, are manufactured from substances extracted or derived from whole foods and typically contain additives, flavorings, and industrial ingredients rarely used in home cooking. Public health researchers have linked high ultra-processed food intake to obesity, cardiovascular disease, type 2 diabetes, and several cancers, prompting intense scientific debate about whether the degree of processing itself, rather than the nutritional profile of the foods, drives disease risk.

The research team, led by investigators at the National Cancer Institute and collaborating institutions, recognized a fundamental complication in this debate. Ultra-processed foods are not nutritionally monolithic. Some, such as whole grain breads and dairy products like yogurt and milk, contain components consistently associated with a lower risk of colorectal cancer, including fiber, calcium, and beneficial whole grain compounds. Others, notably processed meats such as sausages, bacon, and deli slices, contain substances classified as probable human carcinogens and are among the strongest dietary risk factors for colorectal cancer. This internal heterogeneity, the researchers hypothesized, could explain why previous studies examining ultra-processed foods as a single category have produced inconsistent results when it comes to colorectal cancer.

To disentangle these effects, the team analyzed data from 461,682 cancer-free participants aged 50 to 71 who enrolled in the NIH-AARP Diet and Health Study between 1995 and 1996. Each participant completed a detailed food frequency questionnaire, and the researchers performed an unusually granular task: they disaggregated every mixed food item in the questionnaire and assigned its constituent ingredients to Nova processing categories. This approach allowed them to calculate not only total ultra-processed food intake but also the specific amounts of whole grains, fiber, dairy, calcium, and meat consumed from ultra-processed versus unprocessed or minimally processed sources. Median intake among participants was 293 grams per 1,000 kilocalories per day, equivalent to roughly 43 percent of daily energy.

Participants were followed through 2018, a surveillance window spanning more than two decades, during which 10,075 cases of colorectal adenocarcinoma were identified through linkage with cancer registries. The researchers used multivariable-adjusted Cox proportional hazards regression, a statistical framework that estimates how the rate of new cancer diagnoses varies across levels of dietary exposure while accounting for competing influences. Their models adjusted for age, sex, race and ethnicity, education, physical activity, smoking status, and alcohol consumption, among other demographic, lifestyle, and health-related factors known to influence colorectal cancer risk.

The headline result was striking in its nullity. When participants in the highest quintile of ultra-processed food intake were compared with those in the lowest, the hazard ratio for colorectal cancer incidence was 0.97, with a 95 percent confidence interval spanning 0.91 to 1.03 and a trend test yielding a p-value of 0.55. In plain terms, older American adults who ate the most ultra-processed food did not develop colorectal cancer at a measurably different rate than those who ate the least, once other risk factors were taken into account. For a category of foods that has become a lightning rod in nutrition science and public health policy, the absence of an association with one of the most diet-sensitive cancers is itself a consequential finding.

Beneath that null result, however, the component-level analysis revealed a more textured picture. Whole grains, fiber, dairy, and calcium consumed from ultra-processed sources were each inversely associated with colorectal cancer incidence, meaning higher intakes corresponded to lower risk. This suggests that when beneficial nutrients arrive inside industrially processed products, such as fortified breads, cereals, or dairy foods, they appear to retain their protective character. By contrast, meat consumed from ultra-processed sources was positively associated with colorectal cancer incidence, consistent with a large body of evidence implicating processed meats as a causal dietary risk factor for this disease.

These opposing signals within the ultra-processed category likely explain the overall null association. If some ultra-processed foods modestly raise colorectal cancer risk while others modestly lower it, the net effect of total intake may average out to nothing detectable. The authors argue that this heterogeneity is precisely why treating ultra-processed foods as a single monolithic exposure can obscure meaningful dietary relationships. From a prevention standpoint, the findings reinforce existing dietary guidance rather than rewriting it: consume adequate whole grains, fiber, dairy products, and calcium, and limit processed meat, regardless of the processing level of the foods delivering those nutrients.

The study’s observational design imposes important caveats. Dietary intake was self-reported through a food frequency questionnaire, which is vulnerable to both systematic and random measurement error, and the researchers acknowledge the possibility of residual confounding from factors not captured in their models. Participants in the NIH-AARP cohort are predominantly older, non-Hispanic white, and relatively health-conscious compared with the general population, which may limit how broadly the findings generalize. The Nova classification itself, while increasingly influential, involves judgment calls when assigning mixed dishes and ambiguous products to processing categories, and misclassification of exposure remains an inherent limitation of any such scheme.

Even with those constraints, the analysis stands among the largest and most granular investigations of food processing and colorectal cancer conducted to date, and its methodological approach of disaggregating mixed foods into ingredient-level processing categories offers a template for future research on other cancer sites and chronic diseases. As debates over warning labels, taxation, and dietary guidelines for ultra-processed foods intensify worldwide, the study adds a critical layer of precision: the health consequences of these foods may depend less on the fact of their processing than on what they contain, and blanket conclusions about an entire category may obscure both genuine hazards and unexpected protective components hiding in plain sight on supermarket shelves.

Subject of Research: Associations between ultra-processed food intake and dietary risk factors and colorectal cancer incidence in older US adults

Article Title: Associations of food processing and dietary risk factors with colorectal cancer incidence in the NIH-AARP Diet and Health Study: A cohort study

Article References: Abar, L., O’Connell, C. P., Hong, H. G., Herrick, K. A., Kahle, L., Lerman, J. L., Liao, L. M., Zhang, X., Zhang, X., Zhao, L., Zouiouich, S., Sinha, R., Khandpur, N., Martínez Steele, E., & Loftfield, E. (2026). Associations of food processing and dietary risk factors with colorectal cancer incidence in the NIH-AARP Diet and Health Study: A cohort study. PLOS Medicine, 23(10), e1005251. https://doi.org/10.1371/journal.pmed.1005251

Image Credits: AI Generated

DOI: 10.1371/journal.pmed.1005251

Keywords: ultra-processed foods, colorectal cancer, NIH-AARP Diet and Health Study, Nova classification, processed meat, dietary fiber, whole grains, calcium, dairy, cohort study, cancer prevention, nutrition epidemiology