Study design details have been previously published (5
). Between 1992 and 1994, 1,624 participants were recruited at four clinical centers, aiming for equal representation across glucose tolerance status (normal, impaired glucose tolerance, non–insulin-dependent type 2 diabetes), ethnicity (black, Hispanic, and non-Hispanic white), sex, and age (40–49, 50–59, and 60–69 years). Of 1,087 subjects with normal or impaired glucose tolerance at baseline, 906 (83%) returned to 5-year follow-up, among whom 148 type 2 diabetes cases developed. Low educational attainment was associated with reduced follow-up, which somewhat limits generalizability. Analyses included 862 participants with complete data on exposure, outcome, and key covariates. Data completeness was unrelated to demographic variables except the study center.
At baseline, habitual dietary intake was assessed by using a 1-year, semiquantitative 114-item food frequency interview ascertaining both frequency and serving size. We created 33 food groups based on similarities in food and nutrient composition (6
) that were collapsed to create eight DASH food groups (grains, vegetables, fruits, dairy, meat, nuts/seeds/legumes, fats/oils, and sweets). Adherence to the DASH diet was assessed with an index variable (7
). We additionally distinguished whole grain and low-fat dairy to address the qualitative DASH goals. For each food group, a maximum score of 10 was assigned if the recommended intake was met, whereas lower intakes were scored proportionately. If lower intakes were recommended, reverse scoring was applied, and a score of 0 was applied to intakes ≥200% the upper recommendation. The resulting eight component scores were summed to create the overall DASH adherence score (range 0–80) (7
Anthropometric measures were taken in a standardized manner following the protocol. A 12-sample, insulin enhanced, frequently sampled intravenous glucose tolerance test was conducted, and insulin sensitivity and acute insulin response were assessed using minimal model analysis. Acute insulin response was calculated based on insulin levels through the 8-min blood samples prior to insulin infusion.
At 5-year follow-up, individuals who met the World Health Organization criteria for diabetes on their oral glucose tolerance test or who were taking hypoglycemic medication not previously reported at baseline were considered having incident type 2 diabetes.
Multiple logistic regression analysis was used to assess the relationship between the DASH diet and risk of type 2 diabetes. Parameter estimates and 95% CIs were calculated for DASH tertiles. The test for trend across tertiles used the resulting P
value from the type 3 analysis of effects based on the Wald χ2
test. Previous cross-sectional analyses have indicated a significant interaction between DASH adherence and race with respect to baseline BMI and waist circumference (8
); thus, models were additionally stratified by race/ethnicity into white versus minority (blacks and Hispanics) and a DASH score–by–minority race interaction was tested.