In the United States and other Western countries, diet and nutrition researchers face an urgent imperative: Figure out how to solve the crisis of obesity. About 40 percent of the adults and 19 percent of the children and adolescents in the United States have obesity, according to the Centers for Disease Control and Prevention. More and more of them face the increased risks of suffering from diabetes, cardiovascular disease and countless other negative health effects. This situation looks like a single problem from a population standpoint — one that simple guidelines for balancing calorie consumption and expenditure should be able to solve. Instead, a seeming infinitude of variables influence what each of us eats and how the body responds.
That is: Obesity, like cancer, “is not one disease,” says Elizabeth Mayer-Davis, a professor of nutrition and medicine at the University of North Carolina at Chapel Hill. In order to treat it, “you really have to be thinking about biology and behavior and society and culture and policy all at the same time. Because if you miss any one of those pieces, your intervention or your diet — it’s less likely to actually work.” The same diet can affect even identical twins differently. “It’s also why there have been so many conflicting studies in nutrition,” Mayer-Davis says. “The public is very frustrated.” Indeed, just last month a paper in Annals of Internal Medicine created controversy when it argued that there’s not enough evidence to say whether red and processed meats are bad for us, despite years of guidance claiming just that. It also reignited a growing debate: How valuable can universal diet guidelines be for individuals?
In recent decades, popular weight-loss plans have largely seesawed between low-fat strategies, which U.S. health agencies have also promoted, and low-carbohydrate ones. Many of them appear to work especially well for some people and not well for others; on average, however, in studies comparing the two kinds of regimens, participants lose the same moderate amount of weight. In those cases when opposing diets produce equivalent results, Kevin Hall, a researcher at the National Institutes of Health, wondered if there was, in fact, an explanation other than the nutrients. He noticed that many of those diets tended to have at least one rule in common: Avoid ultraprocessed food, the sort of packaged fare containing artificial flavorings and ingredients you wouldn’t find in your kitchen that make processed food cheap, convenient, tasty and shelf-stable — and popular. It currently accounts for 57 percent of the American diet (a proportion that is rising). Previous studies have found correlations between ultraprocessed-food consumption and obesity but no proof that it’s a cause.
In order to look for factors associated with ultraprocessed food that might cause weight gain, Hall recruited 20 adults between the ages of 18 and 45 to move into an inpatient ward at the N.I.H. Clinical Center for four weeks. They were each randomly assigned to one of two groups. One ate meals consisting primarily of ultraprocessed foods, including many that people typically consider healthy: Honey Nut Cheerios, Yoplait yogurt and precooked frozen eggs. The other group ate mostly unprocessed foods, including oatmeal, roast beef, Greek yogurt, fresh scrambled eggs and barley. The meals offered to each group contained an equivalent number of calories and proportions of carbohydrates, fat and sugar; participants ate as much as they wanted. After two weeks, the groups switched diets.
On the ultraprocessed diet, the subjects on average consumed 500 more calories a day and gained two pounds. A possible reason: The participants’ levels of PYY, a hormone that suppresses appetite, were lower on it, while the levels of ghrelin, a hunger-stimulating hormone, were higher. Because the nutrients were constant, the researchers theorize, the processing itself may somehow trigger the hormonal changes. In a paper published in Cell Metabolism in May, they write that “limiting consumption of ultraprocessed food may be an effective strategy for obesity prevention and treatment.”
The results seemed to reinforce a general, even obvious consensus among nutrition researchers — and yet they generated criticism anyway, much of which highlights fundamental challenges in designing dietary experiments. One charge leveled against Hall’s study, for example: It was too short to observe metabolic and behavioral changes that often take place more than two weeks after a new diet is started. But as Hall points out, it would have been hard to recruit people to live longer under such strict conditions. What’s more, outside the lab, subjects often fail to stick to a diet precisely, and their habits are so variable that it can be tricky to tell for sure whether the meal plan in question is having an effect. Yet it’s impossible to say whether a diet that works in the lab will succeed unless you can study it in real life. “I do think there are ways to get accurate data in free-living people,” Dr. Lydia Bazzano, an epidemiologist at the Tulane University School of Public Health and Tropical Medicine, says. “If you can’t, we’re all sunk.”
At stake is the question of how government agencies and other institutions should deploy limited resources to address a catastrophic national health problem. How should dietary research be conducted in order to ultimately produce results that will be most useful to the public? For instance, most clinical trials are set up, as Hall’s was, to answer an either-or question — say, “Does this diet cause weight loss?” — traditionally by comparing an experimental group with a control. But in nutrition, the answer is often “it depends.” To try to understand how multiple factors might influence a diet’s success, Mayer-Davis and others are working on sequential, multiple assignment, randomized trials (SMARTs), whose subjects, in her words, “move from one treatment to another one to another one over time, depending on how they’re doing.” In a continuing 10-month N.I.H.-funded pilot study, one of the first to use this method to assess dietary strategies, she and colleagues randomly assigned volunteers between 19 and 30 with Type 1 diabetes who are overweight to one of three diets: a low-fat plan or a low-carbohydrate plan, both of which were low in calories, or a Mediterranean-style plan. At three and six months, if they don’t like the diet, haven’t lost 2 percent body weight from their last check-in or are having trouble maintaining their blood-sugar levels, they will move to a different plan. Afterward, statistical analyses will determine what characteristics, including those related to lifestyle, were associated with success, or a lack thereof, for each treatment. Eventually, researchers hope, that sort of information will allow clinicians or even a smartphone app to create a personalized diet.
Of course, even personalized diets would still need to be adhered to, and this raises another quandary. “Fundamentally, especially in the Western world, our diet is fairly unhealthy,” Corby Martin of the Pennington Biomedical Research Center, at Louisiana State University, says. “With really broad strokes we could improve our health by making modifications to our diet that everyone could make.” So why haven’t we? And what are we more likely to follow: a diet that is more personalized than past ones have been or, as in Hall’s formulation, more general?
The answer to that question is most likely “it depends,” as well. But teasing apart cause and effect in subjects who are leading lives filled with variables could inspire practical innovations. As an example, Martin has helped pioneer a way to more accurately measure what study subjects eat by having them use an app to send photographs of their meals to researchers, who reply with recommendations to ensure they are following the diet being studied. Such interactions could help determine whether a lab-tested diet like Hall’s can work in the outside world; similar feedback might also turn out to be inseparable from a given diet’s success once the experiment is over.