Dietary patterns by reduced rank regression predicting changes in obesity indices in a cohort study: Tehran Lipid and Glucose Study.

OBJECTIVE To examine the association between dietary patterns and obesity indices (BMI, WC, WHR) among Tehranian adults in a 6-year follow-up study. METHODS Within frame of a cohort study in Tehran (mean follow up 6.6+/-0.9 years), 141 adults were recruited with: two 24 hour dietary recalls at the beginning, as well as obesity indices at the beginning and end of the study period. Dietary intakes were converted into grams of intakes of food items and categorized into 16 groups. Reduced rank regression analysis derived five patterns with total and polyunsaturated-to-saturated fat intake, cholesterol, fiber and calcium intake as response variables. Factors (dietary patterns) were generated retaining a corresponding factor loading > or = |0.17| on the food groups. Changes in obesity indices were scrutinized within quintiles of factor scores. RESULTS There were high loadings on refined carbohydrates, whole grain, starchy vegetables, other vegetables, red and refined meat, saturated/trans fat, and egg for the first factor named "traditional". All obesity indices had increasing trend across quintiles of pattern score. The fifth pattern (namely egg pattern) had high loading for eggs, salty snacks, as well as fruits and dry fruits, and negative loadings for red and processed meat, saturated and trans fat, plant oils, and dairy products. This pattern showed increasing trends for WC and WHR after adjustment for potential confounders. Other patterns showed non-significant trends for obesity indices. CONCLUSIONS The results were indicative of a traditional pattern which is dominated in the Tehran region and associated with increase in obesity indices.