These foods have been linked to obesity and other health troubles, but some scientists argue the grouping is too broad to guide dietary choices

Are ultra-processed foods really so unhealthy? What the science says

These foods have been linked to obesity and other health troubles, but some scientists argue the grouping is too broad to guide dietary choices

In the early 2000s, Brazilian nutrition researcher Carlos Monteiro made a puzzling discovery that led to an epiphany. While trawling survey data on household spending to try to understand why rates of obesity and type 2 diabetes were rising so rapidly in his home country, he was surprised to note that people were buying smaller quantities of sugar, salt and other ingredients generally associated with these conditions than they had in previous decades.

Only when Monteiro and his colleagues dug deeper did they find the culprit. People were buying less sugar to prepare cakes and desserts, but eating more of it in pre-made pastries and breakfast cereal. They were buying less salt, but consuming more of it in frozen pizzas, chicken nuggets and dehydrated packet soups. “We realized the problem was our traditional dietary patterns were being replaced by foods that are processed so many times that they can no longer be recognized in the final products. We called them ultra-processed foods.”

Monteiro, a nutrition and public-health researcher at the University of São Paulo, first used the term ultra-processed food (UPF) in a paper in 2009, arguing that people interested in promoting healthy diets should focus more on the degree, extent and purpose of processing than on nutrient profiles¹. It was a radical idea that caught the attention of other researchers, who, over the next decade or so, published dozens of papers linking UPFs with obesity and a range of other health problems.

Governments took notice, too. In 2014, Brazil began advising people to avoid UPFs. Other countries, including France, Belgium and Israel, followed suit. Robert F. Kennedy Jr, secretary of the US Department of Health and Human Services (HHS), has been a critic of UPFs, saying in January that they are “poisoning the American people”. In May, the US government announced plans for a research agenda to support nutrition policy and improve people’s diets, in part by improving understanding of the impacts of UPFs on health.

Although there’s good evidence that diets high in UPFs are associated with poor health, some scientists are sceptical that the UPF category is useful for research or as the basis for dietary advice, arguing that it is too broad and ill-defined. It makes little sense, they say, to lump together foods as diverse as shop-bought yogurts and wholegrain breads with doughnuts and potato crisps. And they wonder whether UPFs are largely linked to poor health just because they tend to be high in fat and sugar, and low in fibre and vitamins. Questions remain, but recent research findings can provide some guidance for people trying to make healthy choices.

Changing diets

For hundreds of thousands of years, humans have been processing food by cooking it. The fermentation, milling, salting, pickling and smoking of food also have long histories. Canning, pasteurization and the use of additives, such as artificial sweeteners and colouring agents, date back to the nineteenth century. Today, many foods undergo multiple forms of advanced processing, such as fractionation into constituent sugars, oils and fibres; hydrogenation to alter physical properties; and extrusion, which subjects ingredients to high temperature, pressure, shear forces and rapid expansion. These foods are generally classified as UPFs.

UPF consumption has grown since the 1950s, initially in high-income countries and, since the 1990s, in low- and middle-income countries². Surveys suggest that UPFs make up almost 60% of calorie intake in the United States and United Kingdom, 48% in Canada and 42% in Australia (see ‘Ultra-processed intake’). Traditional eating patterns have proved more enduring in Romania (15% of calorie intake), Italy (18%), Brazil (22%) and France (31%).

Research suggests that the rise in UPF consumption might be contributing to the increased prevalence of a range of health problems in many countries. Dozens of studies have linked diets high in UPFs with increased risks of obesity, cardiovascular disease, type 2 diabetes, fatty liver disease, cancer, asthma, depression and anxiety³^–⁵.

In a study published last year, for example, epidemiologist Mingyang Song and his colleagues at Harvard University in Boston, Massachusetts, analysed data on the diets and health of more than 110,000 US adults, who were tracked over a median period of more than 30 years. Song found that those with the highest proportion of calories from UPFs were 4% more likely to die of any cause during the study than were those with the lowest intake. Higher UPF consumption was associated with increased mortality from neurodegenerative diseases, but not from cardiovascular or respiratory disease or from cancer⁴. Diets that included more processed meat, poultry, seafood and sugary drinks were associated with higher risks after researchers controlled for factors such as body mass index, physical activity, alcohol consumption and family health history.

The study also suggested that nutritional quality could be a big factor in why diets high in UPFs might be unhealthy. The increase in risk was much lower for those with diets that, although high in UPFs, were still defined as healthy according to a diet-scoring system called the Alternative Healthy Eating Index. Those diets included a lot of vegetables and nuts, and limited the intake of sugary drinks, red and processed meats, salt and trans fats. “The association between UPF consumption and mortality was much weaker once we accounted for dietary quality,” says Song.

Still, other research indicates that nutritional composition is not the whole story. When Samuel Dicken, a clinical scientist at University College London, reviewed 37 studies that reported health outcomes linked with UPFs, he found that almost all of the associations remained after researchers controlled for nutritional composition⁶.

Dicken and his colleagues subsequently ran a trial in which adults who were overweight or obese were asked to follow one of two diets defined as healthy according to the UK Eatwell Guide (see go.nature.com/4fpdeen), and then switched to the other diet. One of the diets was made up mostly of UPFs and the other mostly of minimally processed foods, but the two diets were matched for levels of protein, carbohydrate, fat, fibre, sodium and sugar. Participants lost twice as much weight (an average of 2 kilograms versus 1 kg) on the minimally processed diet compared with the UPF diet⁷. The findings, published in August, suggest that the increased prevalence of health problems linked to a UPF diet in other studies cannot be fully explained by poor nutritional composition. So what other mechanisms might be involved?

Beyond processing

Dicken’s study is one of only a handful of randomized controlled trials (RCTs) on UPFs. Unlike retrospective and observational studies, RCTs can demonstrate causal links and so might offer a clearer picture, researchers say.

The first RCT on UPFs was carried out by a team led by integrative physiologist Kevin Hall, who until April was at the US National Institutes of Health (NIH), part of the HHS. For more than a decade, Hall has sought to overcome some of the limitations of nutrition research by confining study participants to a research hospital, so that their diets can be precisely recorded.

In 2015, he was approached by two Brazilian researchers at a meeting after he presented the results of one such study, comparing a low-carbohydrate, animal-product-based diet with a low-fat vegan diet. “They told me they had enjoyed my talk, but that my focus on nutrients was very twentieth century and that really, I should be thinking about the extent and purpose of processing food,” says Hall, who had not then heard of UPFs. He was sceptical but intrigued enough that he devised a trial in which 20 adults stayed at the NIH Clinical Center in Bethesda, Maryland, and consumed a diet high in either UPFs or unprocessed food for two weeks before swapping diets.

Breaded chicken nuggets are inspected before being packaged at a factory near Orléans in France.Credit: Guillaume Souvant/AFP via Getty

Like many other researchers, Hall defined UPFs using a system developed by Monteiro. Known as NOVA, it classifies food into four groups according to the extent, type and purpose of processing⁸. Unprocessed or minimally processed foods, such as vegetables and pasta, are at one end of the scale. At the other end are UPFs, which undergo multiple industrial processes and usually contain substances rarely used in home kitchens, such as high-fructose corn syrup, hydrogenated oils and additives, including colouring and emulsifiers.

Participants in Hall’s RCT could eat as much as they liked, but the food presented to them was matched for nutrient groups. To Hall’s surprise, the results, published in 2019, showed that participants consumed around 500 calories more per day on average while on the UPF diet — and gained an average of 0.9 kg over the two weeks⁹. They lost that same amount of weight while on the unprocessed diet.

When Hall dug further into the results to try to understand why people on some diets, including those high in UPFs, eat more calories, he and his colleagues found that people consumed more if they ate quicker¹⁰. Participants also consumed more calories when the foods were energy dense (containing more calories per gram) and “hyper-palatable” — containing ingredients high in combinations of fat, carbohydrates, sugar and salt, which, some researchers say, activate reward signals and circumvent satiety mechanisms in the brain¹¹. Energy-dense foods provide more calories in less time than those with fewer calories per gram because they are easier to eat, and many UPFs are energy dense because their water content has been removed.

But questions remained about what might be driving the extra intake, so Hall set up another trial, this time involving 36 people who spent a week each eating a minimally processed diet and three UPF diets: one was energy dense and hyper-palatable; another was energy dense but not hyper-palatable; and the last was neither energy dense nor hyper-palatable.

The study is ongoing, but early findings show that participants ate more and gained more weight while on the energy-dense, hyper-palatable UPF diet compared with the minimally processed diet. Most of the increases in consumption and body weight that occurred on the energy dense, hyper-palatable UPF diet also occurred on the energy dense diet low in hyper-palatable foods, suggesting that energy density could be the more important factor to health.

Previous research has suggested that eating more slowly and eating diets with low energy density are associated with lower calorie consumption — and that various factors, including food texture, can affect eating speed¹².

In one study, Ciarán Forde, a nutrition researcher at Wageningen University & Research in the Netherlands, asked 50 adults to eat as much as they liked during four lunches. He found that they ate much less when presented with hard-textured foods, such as potato waffle fries, than when given soft-textured foods such as mashed potato, regardless of whether the foods were UPFs¹³. “What we saw was that the speed of eating and the texture properties of meals drive consumption, not the degree of processing,” he says.

Forde’s team went on to ask 41 adults to follow, for 14 days, either a UPF diet consisting of foods with harder textures associated with slow eating, or a UPF diet made up of softer foods linked to faster eating, and then swap over. The meals presented were matched for energy density, and, Forde says, the energy density of foods that participants actually consumed turned out to be the same as well. Although the full results are yet to be published, participants consumed 369 fewer calories per day, on average, when their diet consisted of slowly consumed foods than when it contained those linked to faster consumption, according to initial results Forde presented at a meeting of the American Society for Nutrition in Orlando, Florida, in June.

Maybe a combination of energy density and food texture explains why people eat more, says Forde, who has received funding from multinational food companies. “Maybe it’s not something magic in ultra-processed foods that we don’t understand. Rather it’s just about factors we’ve long known, about how foods are eaten and how they drive meal size.”