*2.2. Dietary Information*

Dietary information was extracted from NHANES, which was collected through 24-h dietary recall interviews in the mobile examination center and was validated by the Nutrition Methodology Working Group [32]. Following the DII calculating protocol reported by N. Shivappa et al. [22], 28 food parameters in NHANES were used to calculate the DII, including carbohydrates, protein, total fat, alcohol, fiber, cholesterol, saturated fat, monounsaturated fatty acids (MUFAs), polyunsaturated fatty acids (PUFAs), niacin, vitamin A, thiamin, riboflavin, vitamin B6, vitamin B12, vitamin C, vitamin D, vitamin E, Fe, Mg, Zinc, Selenium, folic acid, β-carotene, caffeine, energy, n-3 fatty acids, and n-6 fatty acids. Previous studies have reported that DII calculated based on less than 30 food parameters kept its predictive ability [33,34].

A lower negative DII score suggests an anti-inflammatory effect, while a higher positive DII score means a pro-inflammatory effect of diet. According to the methods of N. Shivappa et al. [22], the DII calculation should be standardized to a world database that contains standard mean and standard deviation for each food parameter. The database was constructed by examining the relationship between parameters, including food components, and inflammation, in 1943 published articles. A parameter with proof of anti-inflammation effect would obtain a score of "−1", while a food parameter would receive a score of "+1" if it was reported to be correlated with a reduced level of anti-inflammatory cytokines or increased level of proinflammatory cytokines.

These values were further weighted according to the study design. For each included parameter, we first extracted the individualized consumption value and then subtracted it from the standard mean and divided this value by the standard deviation. To minimize the effect of right skewing, these values were converted to a centered percentile score. To achieve a symmetrical distribution with values centered around 0 and bounded between

−1 and +1, each percentile score was doubled, and then we subtracted "1". This centered percentile value for each food parameter was then multiplied by its corresponding inflammatory effect score to obtain the DII score for each food component. Finally, 28 food-specific DII scores were summed to create the overall DII score for each participant.
