People with diabetes have long known that the same food can affect different people’s blood sugar differently. Now, for the first time, there’s scientific proof: Scientists report that a food’s effect on blood sugar depends not just on the food, but on the person eating it.

Blood glucose, commonly known as blood sugar, may be best known for its association with diabetes, but it’s important for everyone to understand. Glucose travels through your bloodstream, delivering energy to every part of your body. Every time you eat, your blood sugar jumps up. Scientists call this jump a postprandial glycemic response (PPGR).

There are two popular methods for predicting how a given food will affect a person’s PPGR: the number of carbs in a food and the glycemic index. Both strategies assume that a food creates the same PPGR response no matter how it’s eaten—or who eats it.

That assumption may be flawed, says a team of researchers from Israel’s Weizmann Institute. Their new report, published last week in the journal Cell, argues that universal guidelines may actually be encouraging people to eat foods that make their blood sugar worse.

“Ascribing a single PPGR to each food … assumes that the response is solely an intrinsic property of the food,” study co-author Eran Segal told The Atlantic. “But there are very striking differences between people's responses to identical meals.” 

Segal and his colleagues recruited 800 healthy volunteers and gave them questionnaires on their eating habits and medical histories. The volunteers provided stool samples so the researchers could check out their gut bacteria. For one week, they tracked their meals and sleep using a mobile app, while a continuous glucose monitor measured their blood sugar. Everyone ate the same breakfast, but beyond that, what they ate was completely up to them. 

Studying people’s eating habits can be tricky, especially when the data is self-reported. People tend to slack or fudge the numbers when tracking their food. That wasn’t a problem for this experiment, Segal told The Atlantic. These volunteers were motivated: “They joined because we explained that we'd be able to tell them which of the foods they normally eat spike their glucose levels. They came because they wanted to know and we said that if they didn't log properly, we wouldn't be able to tell them.”

The results were dramatic, and completely unique to each volunteer. Foods that caused PPRG spikes in one person had little to no effect on another. The data showed that what and how much you eat matters, of course, but just how it matters varies immensely.

These results weren’t limited to high-carb junk food. One middle-aged woman was working hard to stick to a healthy diet that included lots of vegetables, including tomatoes. But data from her glucose monitor showed that her blood sugar spiked each time she ate tomatoes. The good-for-you produce wasn’t good for her at all.

The researchers’ next step was to turn their results into an algorithm. They recruited a fresh round of volunteers and provided each one with two customized meal plans: one “good” and one “bad.” Half of the meal plans came from nutrition experts, and the other half were generated by the algorithm.

Sure enough, the volunteers’ PPRGs improved during the “good” week—even though each person was eating something different. Even their gut bacteria changed for the better. This was true of both the man-made meal plans and those suggested by the computer; in fact, the algorithm's customized recommendations were slightly more effective than those made by the experts.

The researchers hope their results will inspire a new approach to nutrition and weight management. Co-author Eran Elinav said in a press release that the study "really enlightened us on how inaccurate we all were about one of the most basic concepts of our existence, which is how we eat and how we integrate nutrition into our daily life."

Our scientific and cultural approaches to obesity and diabetes may just be “really conceptually wrong,” he said. Scientists and medical professionals believe “we know how to treat these conditions, and it's just that people are not listening and are eating out of control," Segal said, "but maybe people are actually compliant and in many cases we were giving them the wrong advice."

Other researchers believe it may be too early to draw such strong conclusions, and note that Elinav, Segal, and their colleagues never directly compared their results to the glycemic index.

Still, these findings are making waves. The team won’t have any trouble finding volunteers for their next experiment; the wait list currently includes more than 4000 people.