This national cross-sectional study used a Chinese open cohort: the China Health and Nutrition Survey (CHNS) (16). The CHNS accumulated extensive information from 8 to 15 provinces (autonomous regions, municipalities) about community, family, individual, diet, physical examinations, and biochemical examinations and was, therefore, an appropriate tool for researching the Chinese population. All participants provided written consent before the survey. The CHNS was carried out following the Declaration of Helsinki, and its protocol was approved by the Ethics Committee of the University of North Carolina at Chapel Hill and the National Institute for Nutrition and Health, Chinese Center for Disease Control and Prevention (17). Further details of this survey were available in previous publications (18, 19).

The current study acquired data from the 2009 round of the CHNS. The publicly available dataset can be accessed directly on the Internet: https://www.cpc.unc.edu/projects/china. The CHNS 2009 covered 217 communities, including 12,178 individuals from 4,517 households in 9 provinces or autonomous regions with a considerable difference in geography, and it collected fasting blood in participants ≥7 years old to measure biomarkers for the first time. We excluded respondents who had dietary records deficiency (n = 569), aged <18 years at the time of the survey (n = 1,634), those who were pregnant (n = 78) or breastfeeding (n = 67), and those with missing blood assay results (n = 1,410). Additionally, respondents with a history of apoplexy, diabetes, or myocardial infarction (n = 426) were excluded since these conditions may alter dietary and lifestyle habits. We also excluded respondents with incomplete basic information (n = 17) and implausible daily total calorie intake (> 5,000 or < 700 kcal/day, n = 39) (20). Finally, 7,938 individuals were eligible for inclusion (3,737 men and 4,201 women) in the cross-sectional analysis, with ages ranging from 18 to 94 years. The screening process is shown in Figure 1.

The assessment was conducted using a household inventory obtained from households along with three consecutive 24-h dietary recalls obtained from individuals to estimate dietary intake more accurately (21). In each sampling unit, three consecutive days were equally and randomly selected to assess dietary intake. In addition to the inventory of all foods at the start and end of the 3 days, we also recorded the food purchased, grown, or discarded by each household and the number of person-days during this period (22). The usage of oil and condiments (such as salt, sugar, and sauces) was also estimated according to household dietary data (23). Individual records of daily food, mainly including ingredients, amount, meal location, and cooking method for all food consumption (involved in staple foods, side dishes, snacks, fruits, and beverages), were accomplished using food models and picture aids by trained and experienced interviewers (22). A quality check was performed by comparing the daily dietary consumption based on a household inventory with same-day dietary intake based on 24-h recalls (24, 25).

The calculation of total energy intake was based on previous studies (26, 27). Also, consumption of various foods was obtained with the aid of China Food Composition Tables and then further converted to daily dietary intake of CoQ10 using a previous study that provided an overview of CoQ10 contents in various foods (28–30). The participants were categorized into quartiles as Q1, Q2, Q3, and Q4 according to dietary CoQ10 intake levels in this study.

The serum samples were collected as previously described (31). All samples were tested using Hitachi 7,600 automated analyzer (Hitachi Inc., Tokyo, Japan) in the Beijing central laboratory of China-Japan Friendship Hospital, Ministry of Health (medical laboratory accreditation certificate ISO 15189:2007) with strict quality control (31). TC, TG, HDL-C, and LDL-C were detected by an enzymatic method (Kyowa Medex Co., Ltd., Tokyo, Japan). Apolipoprotein B (ApoB) and apolipoprotein A-1 (ApoA1) were detected by diagnostic reagents (Randox Laboratories Ltd., United Kingdom), as well as lipoprotein(а) [Lp(a)] by diagnostic reagents (Denka Seiken Co., Ltd., Japan). Non-HDL-C was calculated as TC minus HDL-C. The diagnostic criteria for dyslipidemia were as follows (32): (1) TC ≥ 6.22 mmol/L (240 mg/dL) was considered as hypercholesterolemia; (2) TG ≥ 2.26 mmol/L (200 mg/dL) was considered as hypertriglyceridemia; (3) HDL-C < 1.04 mmol/L (40 mg/dL) was considered as low HDL-C; (4) LDL-C ≥ 4.14 mmol/L (160 mg/dL) was considered as high LDL-C.

A portable SECA stadiometer was used to obtain barefoot measurements of height with an accuracy of 0.1 cm, and a calibrated beam scale was used to measure barefoot weight in light clothing with an accuracy of 0.1 kg (31). Weight in kilograms divided by height in meters squared determines body mass index (BMI). The above biochemical and physical measurements were performed uniformly by qualified testers following pre-defined SOPs.

Demographics, health status, and tobacco and alcohol consumption were all collected via an adult questionnaire conducted by qualified investigators. Physical activity levels were assessed according to metabolic equivalent hours per week of occupation, household, leisure, and transportation (26, 33). Urbanization was measured using an urbanization scale with 12 components from the CHNS data (34). For further analysis, the provinces were categorized into the south and the north, bounded by the Qinling Mountain-Huaihe River Line.

Continuous variables were presented as mean ± standard deviation (SD), and categorical variables were presented as n (%). Differences between quartiles were examined with ANOVA for continuous variables and Chi-squared or exact Fisher test for categorical variables. General characteristics were described according to quartiles of dietary CoQ10 intake. Multiple linear regression models were used to investigate how dietary CoQ10 intake influenced lipid profiles (Lp(a), TC, TG, HDL-C, LDL-C, ApoA1, ApoB, and non-HDL-C) and adjusted for age, sex, daily total energy intake, smoking and drinking status, urbanization index, region, physical activity, education, and BMI. The results of linear regression were presented as beta (β) and 95% confidence intervals (CI) using the lowest quartile (Q1) as the reference group. Further, we explored the nonlinear relationships between dietary CoQ10 intake and lipid profiles via the restricted cubic spline (RCS) with 4 knots. Finally, subgroup analyses by sex (male, female), age (< 60 years, ≥ 60 years), and region (north, south) were carried out using the same covariates except the stratification variable.

In addition, we performed a sensitivity analysis to determine the robustness of the findings of multiple linear regression. Energy-adjusted dietary CoQ10 intake, calculated from the mean dietary CoQ10 intake plus the residuals from the regression of dietary CoQ10 intake on daily total energy intake, was used as a substitute for unadjusted dietary CoQ10 intake.

The data were analyzed using R version 4.1.3. p < 0.05 was considered significant.

The general characteristics of 7,938 eligible participants by quartiles of CoQ10 intake from dietary sources were shown in Tables 1, 2 and Supplementary Table S1. The mean dietary CoQ10 intake for all participants was 5.4 mg/day, with a median [interquartile ranges] of 3.92 [1.88, 6.96] mg/day. Meat and meat products were the most important sources of dietary CoQ10, as shown in Supplementary Table S2. Moreover, the median [interquartile ranges] of dietary CoQ10 intake was 4.14 [2.10, 7.16] mg/day for men, 3.73 [1.73, 6.76] mg/day for women, 4.13 [2.06, 7.23] mg/day for people aged <60 years, 3.36 [1.48, 6.14] mg/day for people aged ≥60 years, 5.00 [1.93, 8.61] mg/day for northern people, and 3.54 [1.85, 5.87] mg/day for southern people. As shown in Supplementary Table S3, dietary CoQ10 intake varied significantly by sex, age, education, region, and BMI (all p < 0.05). The average age of participants was 49.9 years. 47.1% of the participants were male, and 26.1% were older persons (people aged ≥60 years). Furthermore, compared with the participants who had lower dietary CoQ10 intake, those with higher dietary CoQ10 intake were more likely to be younger, have a high proportion of males, have lower concentrations of TC, LDL-C, and non-HDL-C, have higher concentrations of TG and ApoA1 and have a higher intake of energy, carbohydrate, fat, and protein.

The results of linear regression analyses in all eligible participants that examined how dietary CoQ10 intake influenced lipid profiles are shown in Table 3. In the adjusted model 2, the quartiles of dietary CoQ10 intake had significant associations with TC, TG, LDL-C, ApoA1, and non-HDL-C (all P _trend < 0.05). To be specific, compared with participants who had the lowest quartile (Q1 < 1.88 mg/day) of dietary CoQ10 intake, those who had the highest quartile (Q4 ≥ 6.96 mg/day) showed significantly higher levels of TG (β = 0.14 mmol/L, 95% CI: 0.05 to 0.23 mmol/L, p = 0.002) and ApoA1 (β = 0.10 g/L, 95% CI: 0.08 to 0.13 g/L, p < 0.001) and significantly lower levels of TC (β = −0.12 mmol/L, 95% CI: −0.19 to −0.06 mmol/L, p < 0.001), LDL-C (β = −0.17 mmol/L, 95% CI: −0.23 to −0.10 mmol/L, p < 0.001), and non-HDL-C (β = −0.12 mmol/L, 95% CI: −0.18 to −0.05 mmol/L, p < 0.001). There were consistent results when using energy-adjusted dietary CoQ10 intake (Supplementary Table S4).

Figure 2 shows the nonlinear dose–response relationships between dietary CoQ10 intake and lipid profiles. Most of the lipid profiles displayed nonlinear relationships with dietary CoQ10 intake (P _nonlinear < 0.05), except Lp(a) and HDL-C. There was a slight rise of TC, ApoB, and non-HDL-C within the lower level of dietary CoQ10 intake, which peaked at around 3 mg/day of CoQ10 and then reduced with increasing dietary CoQ10 intake. With increased dietary CoQ10 intake, TG and ApoA1 increased continuously and gradually stabilized. A sustained reduction was observed for LDL-C with increased dietary CoQ10 intake and then stabilized.

Subgroup analyses compared participants stratified by sex, age, and region (Tables 4–6). The associations between dietary CoQ10 intake and Lp(a), TC, HDL-C, LDL-C, ApoA1, ApoB, and non-HDL-C showed similar trends in different subgroups. In addition, the positive associations between dietary CoQ10 intake and TG were considered statistically significant only among women (Q4: β = 0.16 mmol/L, 95% CI: 0.05 to 0.26 mmol/L, P _trend = 0.006), participants aged <60 years (Q4: β = 0.18 mmol/L, 95% CI: 0.07 to 0.29 mmol/L, P _trend = 0.003), and participants living in the north of China (Q4: β = 0.26 mmol/L, 95% CI: 0.12 to 0.41 mmol/L, P _trend < 0.001).