The theory supporting low-carb dieting lacks evidence. But if it leads to healthier eating overall, does it matter?

The Low-Carb Lowdown

In his new book “The Case Against Sugar,” journalist Gary Taubes makes, as you might easily guess, a spirited case against sugar. His argument is based on the straightforward idea that sugar contributes to obesity and disease well beyond its calorie content, because it affects human metabolism in a way that encourages fat storage.

In his new book, the science journalist Gary Taubes takes a hard-nosed look at sugar — and further advances the idea that not all calories are created equal. But there are competing theories of obesity. Who’s right?

Sugar is a carbohydrate, and the body converts carbs to glucose which is then absorbed into the bloodstream. This, in turn, triggers the pancreas to release insulin, the hormone that enables the body to use energy or store it as fat. If a person doesn’t eat many carbohydrates, the pancreas doesn’t release as much insulin, and less fat is stored, forcing the body’s metabolism to increase and burn off that energy. In practical terms, the theory goes, such a person will have an easier time losing weight — or avoiding gaining it. This hypothesis is called, appropriately, the carbohydrate/insulin, or C/I, model, and it is the basis for any number of popular low-carb diets, including Atkins, the Paleo diet, and others.

It is also a “minority position” among food scientists, Taubes concedes, and many mainstream nutrition authorities reject it.

The debate is often framed as being over the nature of calories themselves, with scientists holding that calories are units of energy — each one no different than the other — and that obesity and related health problems are a result of a simple energy imbalance: More calories are coming into the body than are, for whatever reason, being burned. Taubes dismisses this notion as “inane,” but the disagreement isn’t really about calories at all. It’s about competing theories of obesity.

Is it caused, as Taubes contends, by an excess of insulin, driving energy into fat storage? Or is it caused by an excess of calories, which the body is forced to process and store as fat? No one really disputes that to avoid gaining weight a person must burn or otherwise expend all the calories ingested each day. The question is whether some ideal combination of calories from specific sources — fats, carbohydrates, and proteins — can prompt the body to burn more without your ever having to set foot on a treadmill.

The carbohydrate/insulin model posits that there is — and the great thing about that model is that it’s testable. Do low-carbohydrate diets reduce insulin?  We can measure that. Do low levels of insulin decrease fat accumulation, compared to higher levels?  We can measure that too, at least in theory. In his book, Taubes details historical and epidemiological evidence for carbohydrates’ unique role in obesity, but little lab testing has been done in humans.

Taubes wanted to change that. Back in 2012, he co-founded a nonprofit organization called the Nutrition Science Initiative dedicated to fighting obesity and its related diseases by improving the quality of nutrition research. This included testing the C/I model, and NuSI assembled a team of very well-respected nutrition researchers to do it.

The team certainly wasn’t the first to tackle the problem, but few previous studies were as rigorous and specific as the ones NuSI set out to do.  For a study to be comparable, it would require subjects to be confined at the research facility, to ensure that they didn’t succumb to the temptation of outside food; it would have to vary fat and carbohydrate content, but keep protein the same, because protein is already widely understood to increase the body’s metabolism; and it would have to measure insulin levels, fat gain or loss, and overall energy expenditure.

It was a tall and expensive order, and Kevin Hall, a researcher with the National Institutes of Health and a skeptic of the carbohydrate/insulin model, was recruited to spearhead a multi-site study. That was a deliberate choice, according to Hall, because it would lend credibility to the results, which found that subjects obtaining just 5 percent of their calories from carbohydrates saw their insulin levels drop and their energy expenditure rise by an average of 57 calories per day. The expenditure was highest when subjects first went on the diet — which lasted for four weeks — and then decreased over time. It was not, however, accompanied by a fat loss greater than the one the subjects experienced on the control diet.

For all of this, the study didn’t go completely as planned, and not everyone agrees on what the results actually mean. The first phase of the trial kept subjects on a standard American diet, and the goal was to have them neither gain nor lose weight. But the researchers miscalculated the calories required, and subjects lost weight, which might have changed the way they responded to the low-carb diet that followed. “The fact that they couldn’t keep the people in energy balance from the get-go,” says Taubes, “meant the experimental model didn’t work.”

Whether the weight loss made a difference to results — and in what direction — is still debated, but even if the trial were methodologically impeccable, there’s a limited amount that can be learned from one small study. Hall went on to write a paper putting the trial in the context of another he had conducted (with similar results) and related evidence. “There has never been an inpatient controlled feeding study testing the effects of isocaloric diets with equal protein,” Hall wrote, “that has reported significantly increased energy expenditure or greater loss of body fat with lower carbohydrate diets.”

Some studies, in fact, found small decreases in expenditure. Differences in body fat were insignificant, leading Hall to conclude that “important aspects of carbohydrate-insulin model have been experimentally falsified.”

Those are strong words in scientific circles, but Hall’s conclusion was unequivocal: When people are fed carefully controlled diets that vary in fat and carbohydrates, experimenters just don’t see the results predicted by the C/I model. Taubes explains this by pointing out that a very small effect — too small to be experimentally detected, but consistent over decades — is all that’s necessary to wreak health havoc. “If you store 20 calories a day, that’s two pounds in a year,” he says. Eventually, “you’ll go from being lean to being obese.”

He’s right of course, but it’s hard to see how a 20-calorie daily difference matters when calorie consumption in the United States has increased by hundreds of calories per day in recent years. According to the USDA’s proxy for calories consumed, Americans have gone from eating about 2,000 calories a day in 1980, to 2,500 in 2010. In that context, the C/I effect would have to be both significant enough to be responsible for the upsurge in obesity and disease, yet so small that state-of-the-art equipment cannot detect it.

On its face, that seems implausible, but Taubes thinks the stakes are too high to rule out even long-shot explanations for the rapid rise in obesity, diabetes, and related disorders — all of them amounting to a public health crisis. “Let’s say there’s a 10 percent chance the C/I model is right,” he says. “We’re spending a billion dollars a day on this. These are awful disorders.”

It’s a fair point, but not everyone agrees that proving the C/I model right — or wrong — is worth it. “Nitpicking whether you can get slightly better benefits — and, even if you take the best case scenario we’re only talking about slightly better,” says Yoni Freedhoff, a professor of family medicine at the University of Ottawa who studies obesity, “does a disservice to patients by implying that there’s a right way and a wrong way to lose weight.” Even if low-carb diets delivered a small boost in energy burning, Freedhoff suggests, it’s demoralizing if the patient can’t enjoy food on that diet, and irrelevant if she can’t stick to it.

The single biggest determinant of diet success, Freedhoff adds, is whether the patient can turn that diet into a lifestyle change — and that may well be where the C/I model proves most useful, regardless of its overall effect on the body’s calorie management systems. Hall, for example, tells the story of a cardiologist he met with to discuss the results of his trial. “At the end of the day,” the cardiologist told him, “I need a story to tell my patients.” If Americans have been led to believe that the low-fat fad has been debunked, and that carbohydrates are now the enemy, that’s a story, the cardiologist told Hall. “That motivates people.”

In that light, if the C/I proponents find success with that approach and manage to get people to stick to their diets, that’s a positive outcome, regardless of the underlying science.

Still, the question remains: If carbohydrates really were at the root of our problem, and cutting them out the crux of the solution, wouldn’t the millions of people who have subscribed to the Atkins and South Beach and Paleo diets have persuaded the rest of us by now? Not necessarily, says Taubes, who suggests that the ad hoc societal test of the low-carb solution lacks certainty. “If you understand beyond a shadow of a doubt that your disease is caused by sugar and flour and refined carbohydrates,” he says, “you are more likely to adhere to a diet that cuts them out.”

In the end, Taubes concedes that there is no certainty here. He’s also forthcoming in admitting that — like every other journalist or scientist or nutrition expert — he has his biases. Having a large financial and journalistic stake in the matter is one of them.

“Life is better if I’m right,” Taubes says. “I want to put my kids through college, and the more books I sell, the better. If I’m wrong, I gotta find a different career, maybe selling shoes.”

CORRECTION: An earlier version of this article mischaracterized the results of a study conducted by the Nutrition Science Initiative, which examined the effects of carbohydrate-restricted diets compared to control diets. It is not true that study subjects obtaining just 5 percent of their calories from carbohydrates experienced no fat loss. Rather, their fat loss was no greater than those subjects eating a control diet. The story has been updated.

Tamar Haspel is a journalist who has been writing about food and science for nearly two decades. She writes the James Beard Award-winning Washington Post column Unearthed, which covers food supply issues, and contributes to National Geographic, Fortune, and Cooking Light.