2.1. Study population

All of the study population was from NHANES, a cross-sectional study conducted every two years to assess the nutrition and health status of a representative U.S. population (Zipf et al., 2013). Subsamples of the participants were measured with different combinations of environmental chemicals. To extend the scope of exposure chemicals, we used data from NHANES 2005–2012 and all the participants provided informed consent. The survey protocol was approved by the National Center for Health Statistics Research Ethics Review Board. Information such as demographic characteristics, dietary intake, body examination, and biospecimen measurement was recorded comprehensively. The detailed patient selection strategy was illustrated in Fig. S1. In total, 40790 participants took part in NHANES 2005–2012. Due to the study design, a representative subsample was available of the targeted chemicals (n = 10769 left). We excluded those with missing values of the chemical concentration or anthropometric data (including body weight, height, and waist circumference) (n = 1307 excluded). Those who were less than 20 years old, pregnant, previously diagnosed with cancer or malignancy, or underweight (BMI≤18.5) were excluded (n = 3985 excluded). Participants with missing data on covariates were also excluded from the final analysis (n = 664 excluded). The final sample size was 4813 participants.