NHANES is a research program designed to assess the health and nutritional status of adults and children in the U.S. by uniquely combining interviews and physical examinations. All survey protocols were approved by the research ethics review board at the National Center for Health Statistics, and written informed consent was obtained from all respondents prior to participation.

Participants from the NHANES, 2001–2018, totaling 9 survey cycles, were enrolled. Eligible participants should satisfy the following criteria concurrently: (i) adolescents aged between 12 and 19 years old; (2) participants with available first-day dietary interview data. In total, 13,684 adolescents were retained in the final analysis.

C-DII was formulated in a way like DII. Differing from 45 food parameters used in adulthood DII, C-DII simply used 25 of them due to dietary differences between adults and children. The calculation of C-DII rested on 24-h recall dietary nutrient intake, including carbohydrate, protein, fat, alcohol, fiber, cholesterol, saturated fat, monounsaturated fatty acids, polyunsaturated fatty acids, niacin, thiamin, riboflavin, vitamin B12, vitamin B6, iron, magnesium, zinc, selenium, vitamin A, vitamin C, vitamin E, folic acid, and beta carotene. Based on the association with six inflammatory biomarkers (interleukin-1beta, interleukin-4, interleukin-6, interleukin-10, tumor necrosis factor-alpha, and C-reactive protein), each food is assigned a specific nutrient DII score. The validity of C-DII has been basically confirmed, i.e., it is positively associated with markers of chronic inflammation and oxidative stress, and this association is the same whether in adolescents or children [9, 10].

In detail, C-DII was calculated in three steps. Firstly, the difference between individual’s and global’s daily mean intake divided by standard deviation equals a Z score. Secondly, to avoid “right skewing” and achieve the symmetrical distribution, the Z score is converted to a percentile value and doubled before subtracting “1”. Thirdly, the obtained value times each nutrient’s dietary inflammation index score is the DII of each food nutrient. Overall C-DII scores are the sum of each score obtained by previous steps [9, 10].

In this study, eligible adolescents were evenly divided into 3 groups according to C-DII scores: low C-DII group (−4.91 ≤ C-DII ≤ 0.75), medium C-DII group (0.75 < C-DII ≤ 2.33), and high C-DII group (2.33 < C-DII ≤ 4.97) [11].

Prediabetes was defined according to the following criteria: (i) Hemoglobin A1c (HbA1c) levels: 5.7%–6.4%; (ii) fasting glucose levels 100–125 mg/dl; (iii) glucose tolerance levels: 140–199 mg/dl [7, 12]. It can be defined by meeting one of these criteria.

Age, sex, race/ethnicity, poverty income ratio (PIR), and serum cotinine were included. All covariates under study can be accessed/downloaded from the official NHANES website (https://www.cdc.gov/nchs/nhanes/index.htm).

All analyses in this study have incorporated oversampling, clustering, and stratification procedures to estimate the representative statistics of the general U.S. adolescent population given the multistage complex design of NHANES. Continuous variables are represented as weighted mean (95% confidence interval [CI]). Categorical variables are represented as weighted proportions. Wilcoxon rank-sum test for skewed continuous variables and χ2 test for categorical variables were used for group-based comparisons. Overall and subsidiary weighted mean C-DII scores in U.D. adolescents across survey cycles were determined. The nonlinear relationship of C-DII with prediabetes was examined by restricted cubic spline (RCS) analysis. Weighted logistic regression analysis was implemented to assess the association between C-DII and prediabetes, with and without controlling age, sex, race, PIR, serum cotinine.

Two-sided P less than 0.05 was considered statistically significant. Data were analyzed using Stata software version 17 (StataCorp LP, TX, USA) and R programming environment version 4.2.3.