Reproducibility and construct validity of a food frequency questionnaire for assessing dietary intake in rural and urban Asian Indian adults.

Abstract

To evaluate the reproducibility and construct validity of the Madras Diabetes Research Foundation FFQ (MDRF-FFQ) with biomarkers for its use in epidemiological settings in India. The MDRF-FFQ was administered to 500 participants representing rural and urban areas of 10 Indian states, twice at an interval of 12 months. Reproducibility was assessed using intra cluster correlation coefficients (ICC). Construct validity of carbohydrate and fat intake was assessed using baseline serum lipids by regression analysis. Reproducibility as measured by ICC was 0.50-0.77 for saturated fatty acids (SFA) and energy in urban and 0.61-0.72 for protein and SFA in rural areas. The ICC for food groups was 0.53-0.77 for whole grains, fruits and vegetables in urban and 0.50-0.89 for animal foods and whole grains in rural areas. After adjusting for potential confounders, carbohydrate intake was positively associated with serum triglycerides (TG) (β [SE]: +2.3 [0.72] mg/dL; p=0.002) and inversely with high density lipoprotein cholesterol (HDL) (β [SE]:-0.48 [0.12], p<0.001), while dietary fat and SFA (% Energy) were positively associated with HDL, low density lipoprotein (LDL) and total cholesterol and inversely with TG. The MDRF-FFQ can be considered as a reliable and valid tool to measure the long-term dietary exposure in respect of macronutrient intakes in Indian populations despite diverse dietary practices.