Samples were

stored at −20°C until analysis. Maternal blood specimens (approximately 3 mL) were collected in Eppendorf tubes containing 10 mL of 7% ethylenediaminetetraacetic acid as an anticoagulant at the third trimester study visit. Both maternal urine samples and blood specimens were collected on an empty stomach at 8 AM. When the neonates were born, 1 mL umbilical cord blood was collected in Eppendorf tubes using sterilized syringes and stainless steel needles. Cord blood samples were homogenized and immediately stored at −20°C until further analysis. The mercury assays was performed using the Direct Mercury Analyzer 80 (Model DMA-80; Milestone Inc, Milan, Italy). This automatic mercury analyzer requires no sample digestion or pretreatment. The cleaned samples of Epigenetic inhibitor hair, urine, and blood were added to a nickel boat, which was sent into the instruments.

The combustion-atomic absorption spectroscopy procedures were as described elsewhere.13 To ensure the accuracy of the analytical methods and results, quality control steps included daily calibration with verification of a high- and a low-concentration Obeticholic Acid mw standard for each working range, a procedural blank, and certified reference materials as the standard. Mercury recovery was 90-110%, with >95% precision. The Chinese version14 of the NBNA was used to estimate neurobehavioral development of the neonates at 3 days of age. The NBNA contains five sections: behavior (six indexes), passive muscle tone (four indexes), active muscle tone (four indexes), primary reflexes (three indexes), and general assessment (three indexes). Each index had three grades (0, 1, or 2 scores), giving a total maximum score of 40.15 Neonates with a total score

of 40 were considered well developed. Scores <40 were considered abnormal. Trained examiners were blinded with regard to exposure status when the NBNA was implemented. Descriptive statistics were calculated for each variable, including maternal, paternal, and neonate characteristics. Spearman correlations were used to assess correlations among maternal hair, urinary, and blood mercury and cord blood mercury. Comparisons of the cord blood mercury between each group with different the frequencies of fish consumption were performed using analysis of variance. Predictors of total NBNA and subsection scores were assessed with stepwise multiple liner or logistic regression models, respectively. Meanwhile, statistical significance of NBNA scores and total mercury levels were analyzed by multiple regression models with adjustment for potential confounding factors, including monthly household income, neonatal sex, head circumference, birth weight, and length. P value of ≤0.05 was considered to demonstrate statistical significance. The study chose “full score” and “not full score” as cutoff points in the logistic regression models to determine the relationship between mercury exposure and neurobehavioral factors.