Development and validation of dietary atherogenic index using common carotid artery-intima-media thickness: A food frequency questionnaire-based longitudinal study in Korean adults.

Dietary patterns based on pathways of diseases such as subclinical atherosclerosis can predict diet-disease associations better than dietary quality indices or data-driven indices. We aimed to develop and validate a novel dietary atherogenic index (DAI) in adults using the common carotid artery-intima media thickness (c-IMT) in Korean adults. Participants aged ≥40 years from the Yangpyeong cohort were enrolled in the development set (n = 1373). Two DAIs were developed using 38 predefined food groups as predictors and c-IMT as response variable, using stepwise linear regression (SLR) and reduced rank regression (RRR) models. Based on the associations between DAI and c-IMT, DAIs were then validated in the total, men, and women, internally in 1000 bootstrap sets, and externally in a set combined with 2 other populations (n = 2476). The DAI-S by SLR and the DAI-R by RRR included 7 anti- and 10 pro-inflammatory food groups and 2 anti- and 6 pro-inflammatory food groups, respectively. Total rice and white root kimchi were pro-inflammatory, and mushrooms, other beverages (mostly traditional beverages), poultry, processed seafood, and pizza/hamburger were anti-inflammatory in both the DAI-S and DAI-R. They were internally valid in the multivariable-adjusted models, showing a significant linear trend in all 3 sets; they were also valid in the bootstrap at 95% confidence interval. In the external validation, only the DAI-S was significantly associated with the c-IMT (P trend = .0004, .0133, and .0212 in total, men, and women, respectively). This study indicated that DAIs, particularly DAI-S, may be useful for assessing the atherogenic potential of food-based total diets.