Ultra-processed food showed no link to colorectal cancer in 461,682 older US adults

Ultra-processed food is blamed for a growing list of diseases. Colorectal cancer, which is cancer of the colon or rectum, is one of the diseases it is most often blamed for. The worry has sharpened as more young adults are diagnosed, and one recent review noted that the global rise in these foods parallels rising rates of early-onset colorectal cancer, raising concern about a causal link.

A large American study now complicates the story. Researchers at the National Cancer Institute tracked a cohort of 461,682 US adults, aged 50 to 71 at the start, for nearly two decades. The people who ate the most ultra-processed food were no more likely to develop colorectal cancer than those who ate the least. The paper, with Leila Abar and Caitlin O’Connell as joint first authors and Erikka Loftfield as last author, was published in PLOS Medicine on October 15.

What the team added was a look inside the category. Ultra-processed food takes in most processed meat, but in this study it also supplied most of the whole grain people ate, through packaged bread and breakfast cereal. Those foods pulled in opposite directions: the processed meat tracked with more cancer, and the whole grains, fiber, dairy and calcium tracked with less.

Why ultra-processed food is suspected in colorectal cancer

The term comes from Nova, a scheme that sorts foods by the extent and purpose of industrial processing. Its authors describe these products as built from substances taken out of whole foods or chemically changed, with frequent use of cosmetic additives such as flavors, colors and emulsifiers. They are designed to be cheap, convenient and easy to overeat. Such foods now contribute 57% of total daily calories for American adults, according to one US cohort study.

There are plausible reasons to suspect them. Diets heavy in these products tend to be high in added sugar, refined starch and fat, and they go with putting on weight. Obesity, in turn, is an established risk factor for colorectal cancer. Some additives have drawn attention too: the new paper’s authors note that emulsifiers have been found to disrupt the lining of the gut, a change linked to inflammation and to colorectal cancer.

Not everything in the category points the same way. The authors cite a 2018 report from the World Cancer Research Fund. It linked whole grain, fiber, dairy, and calcium intake with lower colorectal cancer risk, and processed meat, red meat and alcohol with higher risk. Processed meat carries the strongest verdict of all: the World Health Organization classes it in Group 1, and its question-and-answer page states that eating processed meat causes colorectal cancer. Under Nova, a packet of sausages and a loaf of whole-wheat bread can both count as ultra-processed. The National Cancer Institute team set out to see what happens when those two kinds of food are counted separately.

How the NIH-AARP study measured ultra-processed food

The NIH-AARP Diet and Health Study, named for the National Institutes of Health and for AARP, the US membership group for people over 50, began in 1995 and 1996. Researchers sent questionnaires to AARP members aged 50 to 71 who lived in one of six states, or in Atlanta or Detroit. Participants reported their usual diet on a food frequency questionnaire covering 124 food and beverage items, along with their smoking, exercise, schooling and health.

The researchers broke every item on that form into the thousands of specific foods it stands for in national US food databases. Then they assigned each one a Nova group. That let them estimate not only how much ultra-processed food people ate, but how much of their whole grain, fiber, dairy, calcium and meat came from ultra-processed sources and how much from less-processed ones. At the median, ultra-processed food made up 43% of daily energy intake, or calories, and the middle half of participants got between 36% and 50% of their calories from it.

Cancer diagnoses came from state cancer registries, which earlier checks found captured more than 90% of new cases, and deaths from the National Death Index. With a median follow-up of 18.8 years, the researchers counted 10,075 colorectal cancers. Of these, 4,787 were in the first part of the colon, 2,633 in the last part and 2,389 in the rectum; the rest were at overlapping or unspecified sites.

They then compared people in five equal groups, from the fifth who ate the least ultra-processed food (measured by weight, per calorie eaten) to the fifth who ate the most. The comparison took account of age, sex, race and ethnicity, education, smoking, physical activity, alcohol, family history of cancer and how people rated their own health. Separate versions also took account of overall diet quality or body weight.

This is an observational study, so it can show which diets went with more or fewer cancers but not that a diet caused them. The two ends of the scale were not alike. The heaviest consumers were younger and about twice as likely to have obesity (29.6% against 15.2%). They were also less likely to be college graduates or to exercise three or more times a week. Statistical adjustment can account for differences that were measured, but not for those that were not.

What the study found for ultra-processed food and colorectal cancer

People in the top fifth of intake had a colorectal cancer rate about 3% lower than people in the bottom fifth. That gap is small enough to be chance, and the researchers describe it as little evidence of an association. The range of plausible values ran from 9% lower to 3% higher. That is narrow for a nutrition study, and it makes a large increase in risk unlikely to have been missed at the intake levels seen in this group. The result held for each part of the bowel, the colon and the rectum, and it barely moved when diet quality or body weight was added to the comparison.

The picture changed when the category was broken into the foods within it. Eating more of several ultra-processed groups that supply calcium, dairy, fiber or whole grains went with lower rates: breads, breakfast cereals, and frozen desserts and milk drinks. Those who ate the most breakfast cereal had a rate about 16% lower than those who ate the least. Higher intake of reconstituted meat and fish products, a group that takes in many processed meats, went with a rate about 10% higher.

Much of what is thought to protect against colorectal cancer, and much of what is thought to raise it, came from ultra-processed foods. In this cohort, 82.8% of whole grain intake was ultra-processed, along with roughly a third of fiber and calcium, nearly a quarter of dairy, and 74.3% of processed meat intake.

The team also split each nutrient by its source. Dairy and calcium from both kinds of food were independently associated with lower risk, and meat went with higher risk from either source. Each daily serving of dairy from ultra-processed foods went with about 25% lower risk, and each daily serving of meat from them with about 8% higher risk, though, as the limits below explain, these food-by-food estimates are exploratory. Whole grain and fiber from ultra-processed foods went with lower risk too. The equivalent figures for whole grain and fiber from less-processed foods were not clear, which the team put down to how few people got their whole grain that way.

The team also ran a statistical swap. They asked what would happen if a tenth of the weight of ultra-processed food in a diet were replaced with minimally processed food, such as fresh fruit and vegetables. That swap was associated with a 2% lower incidence of colorectal cancer, with a range of 1% to 4% lower. The effect is small, and it does not contradict the main result, since the top-versus-bottom comparison could not rule out a difference of a few percent either way. Diet quality mattered as well: poor diet quality was associated with higher cancer rates, whether people ate a little ultra-processed food or a lot.

That is the team’s explanation for the headline result: the lack of an overall link may be explained, in part, by opposing associations of different ultra-processed foods with colorectal cancer, with foods pulling a single total in both directions at once.

How the result compares with earlier ultra-processed food studies

Earlier research has been divided, and the new study does not settle the division so much as offer a reason for it. In three large US cohorts of health professionals, the fifth of men eating the most had a 29% higher risk of colorectal cancer than the fifth eating the least. Among women there was no link with overall intake. That study, published in The BMJ in 2022, also found lower risk in women who ate more yogurt and dairy-based desserts. It echoes the dairy result in the new paper.

The Norwegian Women and Cancer study followed 77,100 women and recorded 1,625 cases. The quarter who ate the most of these foods had a 24% higher risk than the quarter who ate the least, and a 58% higher risk of cancer in the right side of the colon. Pooling 16 cohort studies with more than 2 million participants, a recent analysis in Nutrition Reviews estimated that high intake was modestly associated with increased colorectal cancer risk, about 13% higher. An umbrella review in The BMJ in 2024, which weighed earlier pooled analyses against each other, graded the evidence on these foods across dozens of health outcomes. It put colorectal cancer among the outcomes graded as weak evidence, a lower grade than heart disease deaths and type 2 diabetes received.

The closest match to the new analysis comes from the European Prospective Investigation into Cancer and Nutrition, which follows adults in ten countries. Among about 450,000 of them, replacing 10% of ultra-processed foods with minimally processed food was associated with a lower risk of colon cancer, about 7% lower. The American figure of 2% covered colon and rectal cancer together, so the two are not an exact match.

The National Cancer Institute team suggests the difference lies in what high-ultra-processed diets contain in each place. The Europeans had a higher intake of ultra-processed meat products, an average of 36.2 grams a day against 20.1 grams in the American cohort. They ate less packaged cereal and bread, the foods that supplied most of the Americans’ whole grain. If what people eat inside the category varies from one country to the next, the same label can carry different risks.

What a single food questionnaire from 1995 could not capture

The weaknesses start with the questionnaire, which, like the food databases behind it, was designed to measure nutrients, as the authors point out, and not processing. A person who reported eating yogurt might have eaten plain yogurt, which is not ultra-processed, or a flavored one, which is, and the questionnaire could not tell them apart. Everyone reporting a given food was assigned the national average mix of products, which blurs real differences between people and can push results toward no effect.

Diet was asked about once, at the start, in 1995 and 1996. These foods have become more common since, and the authors say their estimates likely underestimate current intake. They add that the narrow spread of intake between the top and bottom fifths may also have helped produce a null result. The participants were older, mostly white Americans, and the results may not apply to younger or more diverse groups eating today’s food supply.

The study could not measure additives such as emulsifiers, because it had no ingredient-level information. Some influences on cancer risk were not measured at all, such as the use of painkillers like aspirin and ibuprofen. Others, such as income, were measured only roughly. The analyses of individual foods and nutrients involved many comparisons, and the authors describe them as exploratory, since some apparent links will arise by chance.

What the result means for diet and colorectal cancer

Loftfield’s group reads the results as support for advice that already exists. “Our findings support current dietary guidance to consume whole grains, fiber, dairy, and calcium and avoid processed meat,” they wrote, with bowel cancer prevention in mind. They add that their results do not address whether dairy or fiber from an ultra-processed product acts in the body the same way as dairy or fiber from a less-processed one.

A null result on one cancer leaves the stronger evidence on other diseases untouched. The 2024 umbrella review found its most consistent links for heart disease deaths, type 2 diabetes and common mental disorders. What the NIH-AARP analysis adds is a likely reason why colorectal cancer studies keep disagreeing.

How diet fits alongside screening and family history in one person’s bowel cancer risk is assessed by their doctor.

In this large cohort of older Americans, what was in the ultra-processed food tracked with colorectal cancer, while the amount of it did not.