The glycemic index is used to help diabetics and dieters make healthier choices. But how well it works may have to do with what lives in your digestive system.

Scientists increasingly think that what we eat influences which bacteria thrive in our digestive tracks, and the bacteria in our bellies, in turn, influence our health. A new study indicates this is no less true in people with diabetes.

Scientists say the difference in people’s gut bugs may explain why the customary diet used to control or prevent diabetes — more complex carbohydrates, fewer simple carbohydrates — works so much better in some people than in others.

Read More: Why Nutrition Advice Is so Confusing »

The glycemic index is one commonly used way to categorize foods as safe or unsafe for people with diabetics or prediabetes. Many would-be weight losers also use the rating system to differentiate “good” complex carbs from “bad” simple carbs. The system rates foods based on the average amount of glucose they cause the body to produce.

But, the study published today in the journal Cell argues, those averages are almost meaningless. Foods that rate as low-glycemic, such as lentils, may cause spikes in blood sugar in some people while foods we would expect to be sugar bombs, such as ice cream, can fit well into some people’s diets.

In other words, for some, bread may be a “good” carb, while for others it is a “bad” carb.

“This variability is something that can explain some or all of the general failure of the human race to apply a universal diet to address the obesity epidemic,” said Dr. Eran Elinav, Ph.D., a researcher at the Weizmann Institute of Science in Israel, who is one of the study authors.

Elinav and his colleagues fitted 800 participants with continuous glucose monitors to keep tabs on all of the spikes and valleys in their blood sugar over the course of an entire week. The participants did not have prediabetes, but many were at risk. Most were overweight and many were obese.

Participants ate one of four standardized breakfasts each day and wrote down the rest of what they ate.

This produced a flood of data. Sorting through it with the help of a computer, the researchers concluded that gut bacteria were a major factor causing people to metabolize foods differently.

Read More: The Truth About Probiotics and Gut Bacteria »

The researchers used the connections they found to create customized diet plans for a new set of study subjects. After a week on the diets, participants’ post-meal blood sugar improved as much as their cohorts who got standard expert advice.

“The solution basically shifts the focus from giving grades to foods, which is being done today by grading systems like glycemic index and calories, to the individual,” Elinav said.

It’s an appealing prospect.

David Marrero, Ph.D., president of healthcare and education at the American Diabetes Association, thought the approach was an interesting way to account for the variations in how people respond to foods. But most people trying to keep diabetes at bay may not need such high-tech help.

“What really prevents people from developing type 2 diabetes, particularly those who are at risk, is weight reduction,” Marrero said. “Tailoring diets is only going to be effective if it controls caloric intake and results in sustained weight loss.”

While a personalized diet would likely make life more pleasant for participants by identifying times when it might be OK to eat a favorite food that scores a high rating on the glycemic index, it would only help prevent diabetes if it made it easier for patients to stick to it.

Customization could quickly become complexity, which could lead to rationalizing or giving up altogether, Marrero said.

The individualized diets in the study also led to shifts in the populations of several gut bacteria linked to metabolic health.

For those keeping track of the many species researchers have flagged as possibly relevant to weight and blood sugar, the researchers found increases in Roseburia inulinivorans, Eubacterium eligens, Bacteroides vulgatus, Bacteroidetes phylum, and Alistipes putredinis in participants following the customized diets.

All of these have at least one study linking them in higher numbers to better health. Researchers also pointed to decreases in Anaerostipes, for which smaller populations have been linked to better health.

But it would be foolish to run out and buy prebiotics or probiotics in hopes of losing weight or improving blood sugar, according to George Weinstock, Ph.D., who co-leads the prediabetes section of the Human Microbiome Project at the National Institutes of Health and who was not involved in the study.

Gut bacteria do appear to play an important role in type 2 diabetes and obesity, Weinstock said, but scientists aren’t yet able to pinpoint which are the heroes and which are the villains.

“The paper makes some interesting hypotheses, but these now need validation studies,” he said.

Read More: Doctors Finally Begin to Treat Obesity »