Food and diet

Carbohydrates and heart problems

The Daily Telegraph reported that a “diet high in white bread and pasta 'can double chance of heart disease'”. The news comes from a study of 47,000 Italians, which found the link in women but not men.

This large study looked at the volunteers’ diets and followed them up for almost eight years to see who developed coronary heart disease (CHD). CHD is a potentially dangerous fatty build-up in the arteries that supply the heart and can lead to a heart attack. The researchers found that women who ate higher levels of carbohydrates, particularly carbohydrates that cause a rapid rise in blood sugar (known as high-GI carbohydrates, see below), were at increased risk of developing CHD over the next eight years. The study’s main limitation is that it is difficult to rule out the possibility that other factors could have contributed to the effect observed.

This study suggests that avoiding eating too much high-GI carbohydrate may help reduce the risk of heart disease, at least in women. A randomised controlled trial to test this theory would be ideal, but may not be feasible as controlling people’s diet in the long term is likely to be difficult. For now, the best way to maintain good health and potentially reduce the risk of CHD is to follow a balanced diet, take regular exercise and not smoke.

Where did the story come from?

Dr Sabina Sieri and colleagues from the Istituto di Ricovero e Cura a Carattere Scientifico and other research centres in Italy carried out this research. The study was funded by Compagnia di San Paolo, and the Italian research group was supported by the Italian Association for Cancer Research. The study was published in the peer-reviewed journal Archives of Internal Medicine.

The Independent, BBC News,_ Daily Mail_ and Daily Telegraph reported this research and gave reasonably accurate assessments of the study. However, there is some confusion over which foods have a high or low GI value. The study itself (and some news sources) classifies pasta as low GI, but some news sources report that pasta is a high-GI food.

What kind of research was this?

This was part of a large prospective cohort study called the EPICOR Study, which looked at the causes of cardiovascular disease. This most recent analysis looked at the effect of glycaemic index (GI) and glycaemic load (GL). A food’s GI value indicates how much it increases the level of glucose in the blood compared to eating a standard quantity of glucose or white bread. A food with a high GI increases blood glucose more than a food with a low GI. The GL value of food is calculated by multiplying its GI by its carbohydrate content.

The researchers reported that a diet high in carbohydrates raises blood glucose and insulin levels, increases the level of fatty substances called triglycerides in the blood and reduces the levels of “good” cholesterol. These changes would be expected to increase the risk of cardiovascular disease.

This type of observational study is often the best way examine how lifestyle choices affect health outcomes. It is not usually feasible to use study designs that randomly assign people to follow different lifestyles to compare their effects. However, because the compared groups have not been randomly selected, their outcomes may differ due to the influence of confounders (factors other than the one of interest). For this reason, this type of study needs to take any potential confounding factors into account.

What did the research involve?

The researchers analysed data on 44,132 adult volunteers (30,495 women and 13,637 men, aged 35 to 74 years old) who did not have cardiovascular disease at the start of the EPICOR study. They looked at the diet of the volunteers and followed them up for an average of 7.9 years to see who developed coronary heart disease (CHD). They then compared the risk of developing CHD among those with low-GI and low-GL diets with those with high-GI and high-GL diets.

The researchers recruited participants between 1993 and 1998 across Italy. At the start of the study, the volunteers’ diet in the previous year was assessed using three specially devised food questionnaires, which were tailored to the different regions of Italy. The researchers used published GI values where possible and, where this was not possible, they measured the GI of foods directly. They then used these values to estimate the average dietary GI and GL for each volunteer.

The volunteers also had their weight, height and blood pressure measured, completed lifestyle questionnaires and reported whether they took medication for high blood pressure or diabetes. Individuals being treated for diabetes were excluded from the analysis, as were people with information missing about their diet, lifestyle or other factors such as BMI.

Information on cardiovascular disease and deaths was obtained from hospital discharge and mortality databases. Causes of death were assessed using death certificates and medical records. People suspected of having CHD were identified from diagnoses or CHD treatment recorded in their hospital discharge records, or based on their cause of death. Their medical records were checked to verify that they had CHD.

The researchers then looked at the effect of carbohydrate intake, carbohydrate intake from high- and low-GI foods, sugar and starch, and dietary GL and GI. They compared the group of people with the highest carbohydrate intake, highest GL and highest GI diets (top 25%) with those who had the lowest intake (bottom 25%). They looked at men and women separately, and took into account factors that could affect the results, such as age, overall energy intake, body mass index (BMI), fibre intake, high blood pressure, smoking, alcohol use, education and physical activity. The analyses of GI and GL also took into account saturated fat intake.

What were the basic results?

The researchers found that, among the study participants, the main sources of carbohydrates from high-GI foods were bread (60.8%), sugar or honey and jam (9.1%), pizza (5.4%) and rice (3.2%). The main sources of carbohydrates from low-GI foods were pasta (33.3%), fruit (23.5%) and cakes (18.6%).

During the average 7.9 years of follow-up, only 181 of the 44,132 participants could not be traced. During follow-up, there were 463 cases of CHD.

Women who consumed the most carbohydrate (an average of about 338 grams a day) were twice as likely to develop CHD as women who consumed the least carbohydrates (about 234 grams a day) (relative risk [RR] 2.00, 95% confidence interval [CI] 1.16 to 3.43). This link was not seen in men. Similar increases in CHD risk results were found for women whose diets had the highest GL compared to women whose diets had the lowest GL. Again, this link was not found in men.

Women who consumed more carbohydrate in the form of low-GI foods were not at increased risk of CHD compared to those who consumed less. Women who consumed more carbohydrate in the form of high-GI foods (an average of about 201 grams a day) had a 68% greater risk of CHD than those who consumed the least carbohydrate in the form of high-GI foods (about 88 grams a day) (RR 1.68, 95% CI 1.02 to 2.75). However, the link between the highest average dietary GI and risk of CHD was not significant.

There was no significant link between level of starch or sugar intake and CHD risk in women or men.

How did the researchers interpret the results?

The researchers concluded that “high dietary GL and carbohydrate intake from high-GI foods increase the overall risk of CHD in women but not men” in the Italian population they studied.


The results of this study suggest that high-GI foods may increase the risk of coronary heart disease in women. The strengths of this study include its large size, use of a food frequency questionnaire tailored to the food of different regions, prospective monitoring of CHD and low loss to follow-up. There are some points to note:

  • Although food frequency questionnaires are a commonly used way of assessing people’s diets, they have some limitations. The questionnaire relies on people being able to recall how often and how much they ate of specific foods over the past year, which may be difficult to do accurately. In addition, people’s diets over the past year may not fully reflect their diet before this or during follow-up. This could affect results.
  • The authors note that the GI of a food can vary depending on what other foods it is eaten with, and a food frequency questionnaire cannot take this into account.
  • As with all studies of this type, the results may be affected by factors other than the one of interest. These are called confounders. This study took into account a number of potential confounders, which increases the reliability of its results. However, these adjustments may not have completely removed the confounders’ effects, and unknown or unmeasured confounders may also have an effect.
  • Identification of CHD cases at follow-up was based primarily on hospital and death records. It is possible that some cases of CHD will have been missed. Some people either may not have presented to their GP with symptoms yet, or may not yet have been referred by their GP to hospital for further investigation. Additionally, though people with existing CHD were said to have been excluded at the start of the study, it is unclear from the report how such cases were identified, for example by self-report, report in medical records or by investigation. If less stringent methods had been used to identify cases, it is possible that certain individuals were incorrectly included or excluded from the trial.

Overall, this study seems relatively robust and the authors report that other prospective studies have found a link between dietary GL and GI and risk of CHD in women, but not in men. Everyone should aim to eat a healthy balanced diet, and this study suggests that avoiding eating too much high-GI carbohydrate may help reduce risk of heart disease, at least in women. A randomised controlled trial testing this theory would be ideal, but may not be feasible as controlling people’s diets in the long term is likely to be difficult.

NHS Attribution