**3. Results**

#### *3.1. The Distribution of Blood Hg*

The geometric mean (GM) and 95th percentile of blood Hg among all participants were 3.11 μg/<sup>L</sup> and 9.01 μg/L, respectively. The blood Hg levels were significantly higher in males (GM = 3.70 μg/L) than in females (GM = 2.63 μg/L; Table 1). Blood Hg levels increased until the participants were in their 60s. The blood Hg levels also increased as the BMI, alcohol frequency, and household income increased. Fish consumption > 4 times per week was associated with blood Hg levels that were approximately twice as high as those who rarely ate fish (GM = 4.04 μg/<sup>L</sup> vs. GM = 2.17 μg/L). Smoking, alcohol consumption frequency and amount, cooking types, education, marital status, parity, and menopause were also significantly related to blood HG, but herbal medicine had no influence (Supplementary Materials Table S1).


**Table 1.** Blood Hg distributions by demographic variables (μg/L).

Note: GM, geometric mean; P75, 75th percentile; P95, 95th percentile. *a p*-Value obtained from bivariate analysis (SAS Proc SURVEYREG).

#### *3.2. The Distribution of Lipid Profiles and Hepatic Enzymes*

The blood Hg levels were categorized into three groups—low: ≤2.36 μg/L, medium: 2.36 < Hg ≤ 4.07 μg/L, and high: >4.07 μg/L. Approximately 45% of men and 24.6% of women were in the high blood Hg group. Table 2 shows the distribution of each marker across the blood Hg groups. The GMs of LDL, total cholesterol, and TG increased with blood Hg and was highest in the high blood Hg group for both sexes. HDL tended to decrease with increasing blood Hg in all populations, but the trend disappeared after stratifying by sex. The GMs of hepatic enzymes increased with blood Hg in both sexes, and blood Hg had a significant effect on all of the markers, except TG.



< Hg ≤ 4.07 μ g/L, or high: > 4.07 μ g/L. *b p*-Values are the significance of blood Hg levels for each clinical marker, and the regression model included sex, age, BMI, smoking frequency, alcohol frequency, and income.

#### *3.3. Associations between the Blood Hg Levels and Lipid Profiles*

Table 3 shows the GMs of the blood Hg levels and their associations with the lipid profiles. Blood Hg increased until LDL levels reached 'Borderline high'. Total cholesterol consistently increased with blood Hg in females, but the blood Hg levels decreased with a 'High' total cholesterol classification in males. The blood Hg levels significantly di ffered for each lipid profile after being adjusted for sex. However, the significant di fference for HDL disappeared after considering age. According to the definitions for hyperlipidemia (LDL ≥ 130 mg/dL, total cholesterol ≥ 200 mg/dL, or TG ≥ 150 mg/dL), 61.8% of males (*n* = 1710) and 53.9% of females (*n* = 1989) were hyperlipidemia. The blood Hg levels were significantly higher in the hyperlipidemia group (male: 4.03 μg/L, female: 2.83 μg/L) compared to the non-hyperlipidemia group (male: 3.48 μg/L, female: 2.69 μg/L).


**Table 3.** The geometric means of blood Hg and the associations with hyperlipidemia (unit: μg/L).

Note: GM, geometric mean; LDL, low-density lipoprotein; HDL, high-density lipoprotein; TG, triglyceride. *a p*-Value obtained using two-way ANOVA of the lipid profiles and sex. *b p*-Value obtained by ANCOVA adjusted for age. *c* Hyperlipidemia was identified according to the following criteria - LDL ≥ 130, total cholesterol ≥ 200, or TG ≥ 150.

#### *3.4. Association between Blood Hg and the Hepatic Enzymes*

The blood Hg levels were higher in participants who fell outside of the reference range for the hepatic enzymes (Table 4). The levels di ffered significantly by sex, and the significance remained after adjustment for age. According to the criteria for elevated liver enzymes, 24.3% of males (*n* = 671; ALT > 49 U/L, AST ≥ 34 U/L, or GGT ≥ 73) and 14.0 % of females (*n* = 518; ALT > 49 U/L, AST ≥ 34 U/L, or GGT ≥ 38) had elevated liver enzymes. The blood Hg levels were significantly higher in the elevated liver enzymes group (male: 4.36 μg/L, female: 3.25 μg/L) compared to the normal liver enzymes group (male: 3.64 μg/L, female: 2.70 μg/L).


**Table 4.** The geometric means of blood Hg and the associations with elevated liver enzymes (unit: <sup>μ</sup>g/L)**.**

Note: GM, geometric mean; *a p*-Value obtained by two-way ANOVA of the hepatic enzymes and sex. *b p*-Value obtained by ANCOVA adjusted for age. *c* Elevated liver enzymes were identified according to the following criteria - ALT > 49, AST ≥ 34, or GGT ≥ 73 for males, and ALT > 49, AST ≥ 34, or GGT ≥ 38 for females.

#### *3.5. The Risks of Hyperlipidemia and Elevated Liver Enzymes*

Figure 1 shows the number of participants reporting personal medications; 317 (4.91%) reported hyperlipidemia, 1175 (18.2%) reported hypertension, and 539 (8.35%) reported diabetes. One hundred and forty-four subjects took medication for hyperlipidemia and hypertension, both. And 49 subjects indicated that they took medication for hyperlipidemia, hypertension, and diabetes. Increased blood Hg was associated with a 1.105-fold increase in the odds of hyperlipidemia (95% CI: 1.013, 1.208) (Table 5). Thus, an increase of 1 μg/<sup>L</sup> blood Hg was associated with an 11 % risk of hyperlipidemia. The significance of the odds ratio (OR) estimates remained even after adjustment for personal medications related to hyperlipidemia. For those participants reporting hyperlipidemia and diabetes, the blood Hg GM was 3.60 μg/L, and blood Hg was associated with a 1.105-fold increase in the odds of hyperlipidemia (95% CI: 1.013, 1.207). This remained after adjusting for personal medications (hyperlipidemia and diabetes).

**Figure 1.** Frequency of the reported personal medications. Hyperlipidemia-associated diseases were extracted with the terms 'hyperlipidemia', 'dyslipidemia', 'high blood pressure', 'hypertension', and 'diabetes'. Liver diseases were categorized with the terms 'fatty liver', 'hepatitis', 'liver cirrhosis', 'liver disease', and 'elevated liver enzymes'.


**Table 5.** The relationships between blood Hg and hyperlipidemia and elevated liver enzymes.

Note: GM, geometric mean; OR, odds ratio; 95% CI, 95% confidence interval. *a* Personal medication information was included in the model from the self-reported response. *b p*-Value shows the significance of the odds ratio of blood Hg from logistic regression. The unadjusted model included the main effects of sex, age, BMI, smoke, alcohol frequency, fish consumption, and the two-way interaction of sex and alcohol frequency. Each personal medication information was included in the unadjusted model.

Sixty-three (0.98%) participants reported fatty liver, hepatitis, cirrhosis, liver disease, increased hepatic enzymes, or other liver-related diseases. Increased blood Hg induced a 1.345-fold increase in the odds of elevated liver enzymes (95% CI: 1.206, 1.500). Thus, high blood Hg induced a 35% greater odds of elevated liver enzymes. After adjustment for personal medications related to liver diseases, the OR showed a 1.350-fold risk of elevated liver enzymes.

#### *3.6. Relationships between the Lipid Profiles and Hepatic Enzymes across Blood Hg Groups*

The correlations between the lipid profiles and hepatic enzymes in each blood Hg group are presented in Figure 2. In general, the correlation coefficients showed no associations between the hepatic enzymes and lipid profiles, except for TG. The correlation coefficients between log TG and log hepatic enzymes were 0.28 for ALT, 0.14 for AST, and 0.35 for GGT, and did not differ across blood Hg levels.
