Study population
This research data was taken from the Korean National Environmental Health Survey (KoNEHS) III (2015-2017). KoNEHS is conducted every 3 years from 2009 to understand the exposure levels of environmental chemicals, examine influential factors, and continuously investigate the factors of spatiotemporal distribution and changes in the Korean population. The third stage of KoNEHS was stratified into regional administrative and coastal regions and then classified based on the age, sex, and geographical regions socio-economic status. A total of 233 sampling units were randomly selected. After that, 15 households were selected using the systemic extraction method, and about 15 people were surveyed in each sample area [22].
A total number of 3787 individuals (1648 men and 2139 women) with an age of ≥ 18 years underwent the questionnaires, anthropometric measurements, and collected urine and blood to analyze the levels of environmental chemicals. Three hundred eleven were excluded because of missing data for urinary BPA levels (n = 7), one of alanine aminotransferase (ALT) or aspartate aminotransferase (AST) levels (n = 40), and urine creatinine level (n = 1); those with significant alcohol consumption (n = 121; who consumed alcohol 3 times more than a week and 7-9 cups per time in men (n = 112) and who consumed alcohol 3 times more than a week and 5-6 cups per time in women (n = 19)); those with current pregnant women (n = 22); those with hepatitis or hepatic disease (n = 26); and those with AST/ALT ratio > 2 (n = 94). Finally, 3476 participants (1474 men and 2002 women) were included in this analysis (Fig. 1).
Data collection and diagnosis
Demographic and lifestyle characteristics, including age, sex, drinking and smoking status, physical activity, monthly household income, education level, marriage, and medication taking were surveyed through face-to-face interviews. Based on their response, some variables were categorized as follows: education (less than high school graduate, high school graduate, and college graduate or higher), smoking (never, former, and current), physical activity levels (no, moderate, and vigorous), monthly household income (< 2, 2-3, 3-5, and > 5 million Korean won), and marital status (single, married, and divorced/separated). Drinking status was reclassified as never, former, and current according to the questionnaires: did not drink at all (never), had a drinking experience but did not drink at all in the last year (former), and had a drinking experience and drink more than once a week (current).
Participants with diagnosis of hepatitis or hepatic steatosis and who were presently under treatment or taking drugs were regarded as one that has the hepatic disease. Hypertension was defined by self-reporting a history of hypertension or taking antihypertensive medications. Diabetes mellitus (DM) was defined by self-reporting a history of DM or taking anti-diabetic medications. Hyperlipidemia was defined as a self-reported history of hyperlipidemia, use of anti-hyperlipidemic medication, high-density lipoprotein cholesterol ≤ 40 mg/dL, and triglyceride (TG) ≥ 240. Body mass index (BMI) was calculated as the body weight (kg) divided by height squared (m2).
Urinary bisphenol A concentrations
Spot urine samples were collected from participants and stored at 0-4 °C immediately, and subsequently frozen at −20 °C according to the guideline of National Institute of Environmental Research [23]. Urinary BPA were measured using ultra-performance liquid chromatography-mass spectrometry (Xevo TQ-S, Waters, Milford, MA, USA) [24]. Values below the detection limit were divided by the square root of 2.
Laboratory evaluations
Blood samples were collected from participants at the same time with the urine samples. The samples were centrifuged at 3500 rpm to separate the serum, and stored at 0-4 °C. The separated serum was aliquoted and frozen at −20 °C according to the guideline of National Institute of Environmental Research [23]. AST, ALT, and gamma-glutamyl transpeptidase (γ-GTP) were measured in serum samples [25]. The reference values were AST < 34U/L., ALT 10 - 49U/L, and γ-GTP < 73 U/L for men and γ-GTP < 38 U/L for women.
NAFLD evaluations
The hepatic steatosis index (HSI) is an efficient non-invasive biomarker for NAFLD [26]. The variables in the HSI formula were levels of ALT, AST, BMI, sex, and presence of DM. Subjects were categorized into NAFLD and non-NAFLD using the published cut-off value of 36.
$$ \mathrm{HSI}=8\times \frac{\mathrm{ALT}}{\mathrm{AST}} ratio+ BMI\ \left(+2, if\ diabetes\ mellitues;+2, if\ female\right) $$
Since elevated AST, ALT, and γ-GTP were also used for NAFLD assessment [27,28,29], we defined abnormal AST, ALT, and γ-GTP based on the reference value.
Statistical analysis
The average and standard error (continuous variables), and frequency (categorical variables) were provided in non-NAFLD and NAFLD based on the hepatic steatosis index score. Sample weights were included in order to reconstruct the data at the level of the entire population of Korea. The log transformation was performed because the distribution of urinary BPA concentrations was skewed to the right. BPA concentrations were categorized into quartiles based on the weighted sample distribution. The weighted mean (continuous variables) and weighted frequencies (categorical variables) in general characteristics were described by the BPA quartiles. The general characteristics were compared using the t test and an analysis of variance (continuous variable) or χ2 test (categorical variables).
A multivariate logistic regression analysis was performed to evaluate the relationship between urinary BPA levels and NAFLD. Covariates of age, sex, drinking, smoking, physical activity, monthly household income, education, marriage, and urine creatinine were included in the regression model. Because BMI is used in the HSI formula, it is not included in multivariate analysis as an independent variable to avoid collinearity. In this formula, 2 points was added in female to adjust the lower BMI in women compared to men. Also, because DM was an independent risk factor of NAFLD, 2 point was added in the subject with DM [26]. Thus, sex and DM were used as independent variables in multivariate analysis. Data analyses were performed using STATA (version 16.0 StataCorp LP College Station, TX, USA). P values < 0.05 were considered significant.