Food and diet

GI diet 'debunked' claims are misleading

Today, the Mail Online says, “The GI diet debunked: Glycaemic index is irrelevant for most healthy people”, explaining how “it doesn't matter if you eat white or wholewheat bread”.

This is overgeneralised and misleading, so the diet certainly hasn't been "debunked".

Glycaemic index (GI) measures how quickly foods containing carbohydrates raise blood sugar levels in the bloodstream. It’s used in some diets on the basis that foods that raise blood sugar slowly (low-GI) are considered better for you.

This small US study tried mainly obese people on different high- and low-carbohydrate versions of the GI diet for five weeks at a time.

It found that low-GI diets were no better than high-GI diets at reducing certain risk factors for cardiovascular disease and diabetes.

However, the results came from mainly obese adults, a quarter of whom had high blood pressure – so may not necessarily represent “most healthy people”. The very select group involved in this research makes it difficult to generalise the findings to the wider population.

What this trial tells us is that selecting low-GI foods as a way to reduce risk of diabetes and cardiovascular disease might not be any more beneficial than choosing high-GI foods.

This is food for thought for those aiming to reduce disease risk through dietary modifications, and for health professionals advising them.

Where did the story come from?

The study was carried out by researchers from Harvard Medical School and collaborators. It was funded by the (US) National Heart, Lung and Blood Institute; National Institute of Diabetes and Digestive and Kidney Diseases; the Harvard Clinical and Translational Science Center; the National Center for Advancing Translational Science; and the general clinical research center at Brigham and Women’s Hospital.

The study was published on an open -access basis in JAMA, a peer-reviewed medical journal.

The Mail Online got its headline a bit wrong when saying that the results applied to “most healthy people”, as the study had specific eligibility criteria to include people with a BMI over 25, some of whom had high blood pressure. It was also not correct to say that GI diets have been “debunked”, as the results may not be generalisable to the wider population.

What kind of research was this?

This was a randomised crossover trial (RCT) looking at the effect of different diets on cardiovascular disease and diabetes risk factors. The dietary elements of interest were carbohydrate content and GI.

GI is a measure of how quickly foods containing carbohydrates raise blood sugar levels in the bloodstream. High-GI foods cause a short-term spike in blood sugar level, while low-GI foods cause a more prolonged and smaller rise in blood sugar.

Some popular diets advocate the consumption of low-GI foods, based on the assumption that low-GI is healthier than high-GI. However, the researchers point out that the independent benefits of GI on health are uncertain.

An RCT is one of the best methods to isolate the effects of a dietary intervention such as this. Common issues reducing the reliability of RCTs are a lack of compliance to the diet, high levels of people dropping out of the study, or only recruiting small or highly specific numbers of people. Anything less than a couple of hundred is generally considered small. In this RCT, participants were assigned to trial at least two of the different diets, with a wash-out period in between.

What did the research involve?

Researchers recruited 189 overweight people (all had a body mass index (BMI) of 25 or above) and randomly allocated them to follow one of four strictly controlled diets for five weeks.

After this first phase, they were allowed a break to eat what they wanted for two weeks – called a wash-out period. After the wash-out period, they were randomly allocated a second time to a different diet for a further five weeks. 

To be eligible, people had to have a systolic (upper figure) blood pressure of 120 and 159mmHg and diastolic (lower figure) of 70 to 99mmHg. On this basis, some of the people could have had normal blood pressure, some borderline/pre-hypertension, and some high blood pressure (hypertension).

Other eligibility criteria included being aged 30 or above, and being free from diabetes or cardiovascular disease, and not taking medication related to these conditions.

The researchers aimed to ensure that everyone included in the trial went on two different strictly controlled diets for five weeks, with a two-week gap in the middle.

The background diets from which GI was manipulated were healthy dietary patterns established in the Dietary Approaches to Stop Hypertension (DASH) and Optimal Macronutrient Intake to Prevent Heart Disease (OmniHeart). These are diets that, the authors state, are being recommended in dietary guidelines to prevent cardiovascular disease (CVD).

Participants were randomised to one of four different diets:

  • high-GI, high-carbohydrate
  • low-GI, high-carbohydrate
  • high-GI, low-carbohydrate
  • low-GI, low-carbohydrate

All food and drink was provided and controlled by the researchers. The researchers directly monitored how people stuck to each diet through food diaries and the participants making daily visits to a centre, where the researchers directly observed them eating their main meal of the day. 

The main health measurements of interest were risk factors for diabetes and cardiovascular disease, including:

  • Insulin sensitivity. Taken via an oral glucose tolerance test, this shows how the body metabolises carbohydrates – specifically, how sensitive your body is to the effect of insulin. A tendency towards glucose intolerance can be a sign of higher risk of developing diabetes in the future.
  • LDL cholesterol – so-called “bad cholesterol”.
  • HDL cholesterol – so-called “good cholesterol”.
  • Blood fat levels.
  • Systolic blood pressure – the top number in a standard blood pressure measurement representing blood pressure as the heart contracts.

The analysis was restricted to people who had successfully completed the two diets, one after another, with the two-week gap in the middle.

What were the basic results?

Of the 189 randomised to start the trial, 163 completed enough of the study to be included in the final analysis. Compliance to the diets was high. The average BMI was 32 (BMI above 30 is classed as “obese”) – 92% of participants were obese or heavier. Around a quarter of people (26%) were defined as having high blood pressure. The main findings fell into three groups, summarised below.

Low-GI, high-carbohydrate diet, compared with high-GI, high-carbohydrate diet

  • insulin sensitivity worsened by 20%
  • bad cholesterol increased by 6%
  • good cholesterol, blood fat levels and systolic blood pressure were not any different between the groups

Low-GI, low-carbohydrate diet, compared with high-GI, low-carbohydrate diet

  • blood fat levels reduced by 5%
  • all other measures were not different between the groups

Low-GI, low-carbohydrate diet, compared with high-GI, high-carbohydrate diet

  • blood fat levels reduced by 23%
  • all other measures were not different between the groups

The researchers’ main conclusion was that: “In the context of an overall DASH-type diet [a diet to prevent or help people with high blood pressure], using GI to select specific foods may not improve cardiovascular risk factors or insulin resistance.”


This RCT showed that low-GI diets might not reduce risk factors for diabetes and cardiovascular disease in a group of mainly obese adults. All of these adults were free from diabetes or current cardiovascular disease, although a quarter of them had high blood pressure, and some may have had borderline high blood pressure.

As such, the trial’s participants were a specific group. This means that the results may not be relevant to the general population or other subgroups – for example, those who are a healthy weight or have an existing medical condition, such as diabetes.

However, compliance to the dietary interventions was high and the statistics seemed sound, thereby increasing our confidence in the results. If the findings were replicated in other studies, or if this trial had included more participants and/or been longer in duration, we could have some confidence in saying that for this group, the GI diet did not have the expected benefits. However, for example, if any of the effects of GI took longer than five weeks to occur, this study will not have picked them up. 

The authors themselves make the points that GI is only one attribute of carbohydrate-containing foods. They said: “Further, nutrients often cluster. Hence, the effects of GI, if any, might actually result from other nutrients, such as fibre, potassium and polyphenols, which favourably affect health.”

The study achieved a high compliance to the diets, through food diaries and observation. If this was attempted in real life, compliance would be much less. This would mean that any GI effects would probably be even smaller than was found in this study.

For this group of overweight people, the evidence of the GI diet reducing certain risk factors for cardiovascular disease and diabetes is lacking. The diet certainly haven’t been “debunked” for “most healthy people”, as the Mail Online claimed.

NHS Attribution