Diet Characteristics May Influence Diabetes Risk
Diet Characteristics May Influence Diabetes Risk
Wang J, Luben R, Khaw KT, Bingham SA, Wareham NJ, Forouhi NG
Diabetes Care. 2008;31:2120-2125
The EPIC-Norfolk study was a population-based cohort study originally designed to examine the prospective association between diet and cancer; it now includes other endpoints such as diabetes and heart disease. Residents of Norfolk, England aged 40-79 years were invited to attend a health checkup from 1993 to 1997. They also completed a detailed health and lifestyle survey that included a validated 130-item food frequency questionnaire. From the responses, the energy and weight of food intake was estimated, and dietary energy density (DED) was calculated as the available dietary energy per unit weight of foods.
After excluding patients with diabetes, cancer, or cardiovascular disease at baseline, participants were followed through the end of 2005 to ascertain clinical incident diabetes, defined as entrance into a general practice diabetes register, local hospital diabetes register, hospital admission for diabetes-related admissions, and mortality records with coding for diabetes. The investigators used logistic regression to determine whether DED (in quintiles and also continuously) predicted incident diabetes after controlling for other risk factors such as family history of diabetes, body mass index, waist circumference, and physical activity. They also adjusted alternatively for total energy intake and percentage of intake from fat to isolate the impact of DED.
Of the 21,919 patients included in the analysis, 725 (3.3%) developed diabetes over a median follow-up of 10.2 years. At baseline, subjects who developed diabetes consumed a more energy-dense diet than those who did not. The increased risk of diabetes associated with DED was evident in continuous models and when the population was divided into quintiles. For example, each unit of DED increased diabetes risk by 12% after adjustment for age, sex, and body mass index. Further adjustment for occupational status, smoking, physical activity, family history of diabetes, alcohol consumption, and total energy consumption strengthened the association. Compared with the lowest quintile of DED, the risk of diabetes was 58% greater among those at the top quintile in the fully adjusted model. The authors determined that, compared with the highest DED quintile, participants in the lowest group consumed significantly more fruit, vegetables, and alcohol, and less meat, soft drinks, and a lower percentage of energy from fat.
Several clinical trials from around the world have established that lifestyle modification -- improved diet and increased exercise -- is the most effective intervention for reducing diabetes incidence in high-risk individuals. The current study takes that notion several steps further. First, it demonstrates that the overall characteristics of the diet may be more important than the quantity of calories or the proportion of individual nutrients consumed. Second, this was a population-based study in which most participants did not appear to be at high risk, so the reported associations are generalizable to all.
However, although the study controlled for body mass index, it would have been ideal to include other known predictors of diabetes such as fasting glucose, lipids, and hypertension. Nonetheless, these results establish a need for simple patient-friendly calculators of energy density to assist patients with effective diet modification.
Wang J, Luben R, Khaw KT, Bingham SA, Wareham NJ, Forouhi NG
Diabetes Care. 2008;31:2120-2125
Study Summary
The EPIC-Norfolk study was a population-based cohort study originally designed to examine the prospective association between diet and cancer; it now includes other endpoints such as diabetes and heart disease. Residents of Norfolk, England aged 40-79 years were invited to attend a health checkup from 1993 to 1997. They also completed a detailed health and lifestyle survey that included a validated 130-item food frequency questionnaire. From the responses, the energy and weight of food intake was estimated, and dietary energy density (DED) was calculated as the available dietary energy per unit weight of foods.
After excluding patients with diabetes, cancer, or cardiovascular disease at baseline, participants were followed through the end of 2005 to ascertain clinical incident diabetes, defined as entrance into a general practice diabetes register, local hospital diabetes register, hospital admission for diabetes-related admissions, and mortality records with coding for diabetes. The investigators used logistic regression to determine whether DED (in quintiles and also continuously) predicted incident diabetes after controlling for other risk factors such as family history of diabetes, body mass index, waist circumference, and physical activity. They also adjusted alternatively for total energy intake and percentage of intake from fat to isolate the impact of DED.
Of the 21,919 patients included in the analysis, 725 (3.3%) developed diabetes over a median follow-up of 10.2 years. At baseline, subjects who developed diabetes consumed a more energy-dense diet than those who did not. The increased risk of diabetes associated with DED was evident in continuous models and when the population was divided into quintiles. For example, each unit of DED increased diabetes risk by 12% after adjustment for age, sex, and body mass index. Further adjustment for occupational status, smoking, physical activity, family history of diabetes, alcohol consumption, and total energy consumption strengthened the association. Compared with the lowest quintile of DED, the risk of diabetes was 58% greater among those at the top quintile in the fully adjusted model. The authors determined that, compared with the highest DED quintile, participants in the lowest group consumed significantly more fruit, vegetables, and alcohol, and less meat, soft drinks, and a lower percentage of energy from fat.
Viewpoint
Several clinical trials from around the world have established that lifestyle modification -- improved diet and increased exercise -- is the most effective intervention for reducing diabetes incidence in high-risk individuals. The current study takes that notion several steps further. First, it demonstrates that the overall characteristics of the diet may be more important than the quantity of calories or the proportion of individual nutrients consumed. Second, this was a population-based study in which most participants did not appear to be at high risk, so the reported associations are generalizable to all.
However, although the study controlled for body mass index, it would have been ideal to include other known predictors of diabetes such as fasting glucose, lipids, and hypertension. Nonetheless, these results establish a need for simple patient-friendly calculators of energy density to assist patients with effective diet modification.