Ultra-processed, but not all the same: across 461,682 older Americans followed for nearly 19 years, eating more ultra-processed food did not raise bowel cancer rates overall, because the packaged bread, cereal and dairy pulled one way and the processed meat the other

Ultra-processed foods, the industrial products in group 4 of the Nova classification, now supply more than half the calories American adults eat, and they are often blamed for bowel cancer. The evidence is mixed: some cohorts have found higher colorectal cancer rates among heavy eaters of ultra-processed food, others have not. One reason may be that the group is so broad. It takes in hot dogs and soft drinks, but also packaged wholemeal bread, breakfast cereals and flavoured yoghurt, while the long-standing advice for bowel cancer, from the 2018 World Cancer Research Fund report, is to eat more whole grains, fibre and dairy and less processed meat. A study published on 15 October in PLOS Medicine tried to pull those threads apart.

The researchers used the NIH-AARP Diet and Health Study, which recruited adults aged 50 to 71 in 1995 and 1996 from six US states and two cities. After exclusions, 461,682 people who were free of cancer at the start (274,269 men and 187,413 women) were followed through cancer registries until the end of 2018, a median of 18.8 years. In that time 10,075 were diagnosed with colorectal adenocarcinoma. Each person's answers to a 124-item food questionnaire were broken down into individual food codes and sorted by Nova group. The median intake of ultra-processed food was 293 g per 1,000 kcal a day, about 43% of daily energy.

Overall there was no link. After adjusting for age, sex, smoking, alcohol, education, exercise, family history and self-rated health, the fifth of people who ate the most ultra-processed food had a colorectal cancer rate 3% lower than the fifth who ate the least, a difference well within chance (hazard ratio 0.97, 95% CI 0.91 to 1.03). The result held when diet quality, fibre, calcium or body weight were added to the model, and it was much the same for cancers of the right colon, left colon and rectum.

Breaking ultra-processed food into subgroups told a different story. The highest eaters of ultra-processed breads had 9% lower rates than the lowest, of breakfast cereals 16% lower, and of frozen desserts and milk drinks 18% lower. The highest eaters of reconstituted meat and fish products had 10% higher rates. In this group of older Americans, 82.8% of whole grain intake, about a third of fibre and calcium, and 74.3% of processed meat came from ultra-processed foods.

When the team looked at nutrients by source, the familiar pattern came through either way. Per 300 mg of calcium per 1,000 kcal, colorectal cancer rates were 23% lower when the calcium came from ultra-processed food and 10% lower when it came from other food; per serving of dairy, 25% and 10% lower. Each serving of meat went with 8% higher rates from ultra-processed sources and 4% higher from the rest. Whole grain and fibre from ultra-processed food were linked to lower rates; from other sources the estimates were not significant, likely because so little whole grain came from them. Diet quality still mattered: compared with people eating little ultra-processed food and a good diet overall, those with a poor diet had 22% to 28% higher rates in both the second and the top fifth of ultra-processed intake. And in a model that swapped 10% of food weight from ultra-processed to minimally processed food, rates were 2% lower.

The study has real limits. Diet was asked about once, in the mid-1990s, with a questionnaire designed to measure nutrients, not processing, so a plain and a sweetened yoghurt could not be told apart, and today's food supply looks different. The participants were older and 91% white. The authors had no data on additives such as emulsifiers, and some confounding by things like anti-inflammatory drug use is possible. Their reading is that, for bowel cancer, what a food contains seems to matter more than how processed it is, and that the existing advice stands: whole grains, fibre, dairy and calcium, and as little processed meat as possible.