Are Ultra-Processed Foods Killing Us?

Until recently, Guillaume Raineri, a forty-two-year-old man with a bald head and a bushy goatee, worked as an HVAC technician in Gonesse, a small town about ten miles north of Paris. The area lends its name to pain de Gonesse, a bread historically made from wheat that was grown locally, milled with a special process, and fermented slowly to develop flavor. The French élite once savored its crisp yet chewy crust and its tender, subtly sweet crumb. Raineri would occasionally grab a loaf from a boulangerie after work. He doesn’t consider himself a foodie—“but, you know, I’m French,” he told me.

After Raineri’s wife got a job at the National Institutes of Health, in Bethesda, Maryland, they moved to the U.S. The transition was something of a shock. “The food here is different,” he said in a heavy French accent. “Bigger portions. Too much salt. Too much sugar.” He decided to enroll in a paid study at his wife’s new workplace. It was exploring why the American diet, compared with almost any other, causes people to gain weight and develop chronic diseases at such staggering rates. “I wanted to know what is good for my body,” he told me.

In November, for four weeks, Raineri moved into a room that featured a narrow hospital bed, an austere blue recliner, and an exercise bike, which he was supposed to use for an hour a day. “It’s not as bad as it looks,” he said. His wife took to visiting him at the end of her shifts. Once a week, he spent a full twenty-four hours inside a metabolic chamber, a small room that measured how his body used food, air, and water. He was not allowed to go outside unsupervised, owing to the risk that he might sneak a few morsels of unsanctioned food.

Each day at 9 A.M., 1 P.M., and 6 P.M., Raineri was given an enormous meal—about two thousand calories—and instructed to eat as much as he liked. During the first week, he was offered minimally processed foods such as salad, vegetables, and grilled chicken, and he felt great. But, every Friday, researchers changed his diet. He was soon eating calorie-dense, processed foods that, in his words, “just sat in my stomach”: chicken nuggets, fries, peanut-butter-and-jelly sandwiches. He developed heartburn and began to feel bloated, sluggish, and irritable.

A few days before Thanksgiving, I entered the imposing brick building known as the N.I.H. Clinical Center, the world’s largest hospital dedicated to scientific research. I crossed its cavernous atrium, bought a granola bar (organic expeller-pressed canola oil, soy lecithin, soluble tapioca fibre) at an in-house coffee shop, and took a bite in the elevator. Then I followed Emma Grindstaff, a research assistant, to Raineri’s room.

Raineri was sitting in bed, scrolling through his phone in pale-blue pajamas; biometric activity bands were wrapped around his waist, wrist, and ankle. It was almost time for his daily “resting-energy-expenditure test,” to gauge how his metabolism was changing from one diet to the next. Raineri lay down; Grindstaff dimmed the lights and fitted what looked like an astronaut’s helmet around his head. By measuring what Raineri breathed in and out, a machine could approximate how many calories he was burning, and how many of those calories came from carbohydrates versus fat. (Breaking down fat takes more oxygen than breaking down carbs, and research suggests that people metabolize more fat on a less-processed diet.) A monitor estimated that he’d burn around seventeen hundred calories if he lay in bed for the rest of the day.

After the test, Raineri’s extra-large breakfast was rolled in on a cart. Because observation can influence a person’s eating habits, I was asked to leave. (You might skip that extra donut if someone’s watching.) He got to work on a veggie omelette, tater tots, and a jug of milk that contained added fibre.

In the past half century, nutrition scientists have blamed health conditions such as obesity, diabetes, and heart disease on many features of the American diet, including sugary beverages and saturated fat. These factors surely contribute to Americans’ uniquely poor health. But Kevin Hall, the N.I.H. study’s principal investigator, was researching a possible culprit that wasn’t named until the twenty-first century: ultra-processed food. The problem, Hall believed, might have less to do with high levels of sodium or cholesterol than with industrial techniques and chemical modifications. From this perspective, homemade jam on pain de Gonesse would be fine; Smucker’s on Wonder Bread would not, even if it contained less sugar and fat. “The thesis is that we’ve been focussing too strongly on the individual nutritional components of food,” Hall told me. “We’re starting to learn that processing really matters.”

In recent years, dozens of studies have linked ultra-processed fare to health problems such as high blood pressure and heart attacks, and also to some problems that one might not expect: cancer, anxiety, dementia, early death. One analysis found that women who ate the most ultra-processed food were fifty per cent more likely to become depressed than those who ate the least; another found that men who consumed more had substantially higher rates of colon cancer. (Most of these studies controlled for confounding factors such as income, physical activity, and other medical conditions.)

A focus on a food’s level of processing can lead to odd conclusions, however. Julie Hess, a research nutritionist at the U.S. Department of Agriculture, has pointed out that “ultra-processed food” puts canned kidney beans and gummy bears into the same category. Processing also has some benefits. It prevents food from going bad or being contaminated during storage and transport; it allows more people to eat convenient and varied meals, even when particular foods are not in season; and it helps the world feed a growing population. Walter Willett, a Harvard professor who may be the most cited nutrition researcher in the world, argues that studies like Hall’s are “worse than worthless—they’re misleading.” (He prefers to focus on the combinations of foods that people eat over time, and advocates for plant-based whole foods and the Mediterranean diet.)

While Raineri was having breakfast, I went down to a “metabolic kitchen” in the basement, which looked like a chemistry lab in the back of a restaurant. Raineri’s lunch and dinner were already being prepared; chicken breasts sizzled on a stovetop, and the smell of fried potatoes made my stomach growl. “A lot of chefs like to be creative,” Merel Kozlosky, a woman in a blue baseball cap who serves as the kitchen’s director, told me. “What we’re looking for is people who’re meticulous about following instructions.”

Hall and his colleagues had developed exacting protocols so that less-processed meals would closely match ultra-processed meals in terms of nutrients like salt, sugar, protein, and fat. This was meant to isolate the effect of processing. Tomato slices and lettuce leaves sat on a scale, which weighed food to the nearest tenth of a gram; a large stopwatch, for keeping track of cooking times, ticked nearby. Instructions on a clipboard explained how much Pacific Foods vegetable broth to add to soups A1 through E1, whose salt contents ranged from 0.39 grams to 5.61 grams.

I asked a tall, brown-haired cook which diet he most likes to prepare. “Preparing a day’s worth of ultra-processed meals might take an hour,” he said. “Unprocessed meals could take three or four times as long.” He brought his knife down forcefully, cleaving a carrot in two, and continued: “If I’m swamped, I’d rather make the ultra-processed menu. But if I had to pick one to eat for the rest of my life? Unprocessed, no question.”

A central question of the study is whether, consciously or unconsciously, participants eat more when they’re given ultra-processed foods—and, if so, why. This is why participants are offered such immense portions and can stop whenever they want. At one point, Kozlosky pulled a tray out of a commercial refrigerator. The meal looked as though it could feed a family of four: a tub of salad, a bowl of dressing, a container of beans, a cup of salsa, some shredded cheese, a wild-rice blend, and two pitchers of seltzer. After a meal, researchers weigh each dish to see how much has been eaten.

“Is this processed or unprocessed?” I asked.

Kozlosky smiled. “Ultra-processed,” she said. “Lots of participants can’t tell the difference.”

The term “ultra-processed food” was introduced by a Brazilian epidemiologist named Carlos Monteiro. In the early seventies, Monteiro was a primary-care doctor in the Ribeira Valley, an impoverished part of rural Brazil, and he treated many plantation workers with swollen bellies, stunted growth, and exhaustion. He started to think that they needed better food, in larger quantities, more than they needed medicine. He relocated to São Paulo, hoping to study malnutrition. Then he learned that around a million Brazilians were growing obese each year. Strangely, a shrinking number of people were buying ingredients that doctors blamed for the obesity epidemic, such as salt, sugar, and oil. The paradox troubled him.

In the nineties, many nutrition researchers began to turn their focus away from individual nutrients (antioxidants are good, saturated fat is bad) and toward broader dietary patterns. Monteiro developed a theory. Households that bought less salt weren’t eating less salt. They were no longer cooking. A growing share of their meals arrived in a package. “The issue is not food, nor nutrients, so much as processing,” he wrote in a landmark 2009 paper. Novel behavioral and brain-imaging experiments were showing that eating wasn’t always under our conscious control. Monteiro reasoned that something very bad had happened when industrial food systems started churning out cheap, convenient, and tempting foods. He argued that scientists should classify foods by their most unnatural ingredients and by their means of production.

Almost all our food is processed in some way, but it matters how and how much. According to Monteiro’s NOVA Food Classification System, Group 1 foods are unprocessed or minimally processed: nuts, eggs, vegetables, pasta. Group 2 includes everyday culinary ingredients: sugars, oils, butter, salt. Butter and salt your pasta, and you have a Group 3 food: processed, but not automatically unhealthy. But add a jar of RAGÚ Alfredo sauce—with its modified cornstarch, whey-protein concentrate, xanthan gum, and disodium phosphate—and you’re biting into Group 4 ultra-processed fare. The ingredients of a Group 4 meal tend to be created when foods are refined, bleached, hydrogenated, fractionated, or extruded—in other words, when whole foods are broken into components or otherwise chemically modified. If you can’t make it with equipment and ingredients in your home kitchen, it’s probably ultra-processed. (Monteiro’s rubric did not account for industrially farmed crops and livestock, whose use food companies do not necessarily disclose.)

Monteiro’s peers were not immediately convinced. In the five years after his 2009 paper, there were essentially no scientific studies linking food processing to ill health. It wasn’t clear that his rubric had any more validity than the food pyramid, recommended dietary plates, or the nutrition traffic lights that are used in the U.K. But, gradually, scientists started to test his theory. In 2015, Hall, the N.I.H. researcher, attended a conference on obesity and presented research into low-fat and low-carbohydrate diets. After he left the podium, some Brazilian nutritionists approached him. “ ‘That’s a very twentieth-century way of thinking,’ ” he remembers them telling him. “ ‘The problem is ultra-processed food.’ ” The term sounded nonsensical. Nutrition is about nutrients, he thought. What does processing have to do with it?

Hall, who has short salt-and-pepper hair and often wears a lab coat, originally trained as a physicist. He became fascinated with nutrition after learning to model diseases at a Silicon Valley startup; while in a similar role at the N.I.H., he started working in a “metabolic ward” that was being built to study diet and exercise. Some of his early research examined metabolic changes in contestants on NBC’s “The Biggest Loser,” who’d lost drastic amounts of weight. After the Brazilian nutritionists told him about their theory, he designed a trial that he thought would discredit it.