2.1. Data source and study subjects
In this cross-sectional study, we evaluated subpopulations of adults from the NHANES, which is a multipurpose study designed to evaluate Americans' nutritional and physical health. A standard physical examination is conducted at a mobile examination center after a household interview as part of the NHANES, which has been conducted since 19997. The NHANES was conducted by the National Center for Health Statistics (NCHS; Hyattsville, MD, USA). The sampling method of the NHANES was stratified multistage probability, based on the selection of counties, blocks, households, and persons, which accurately represents the civilian, noninstitutionalized US population. Ethics approval was approved by the NCHS Ethics Review Board.
The NHANES measured serum Cu concentrations from 2011 to 2016. For this study, we included subjects from the NHANES 2011–2016 due to the available data of serum Cu. All data were collected by uniformly trained professionals via household interviews, or an investigation conducted in a mobile examination center. The following criteria were used to restrict our study sample: 1) aged ≥ 19 years old; and 2) excluded subjects with missing copper, HF, coronary heart disease (CHD), myocardial infarction (MI), and stroke information. Finally, 5154 participants with complete information were included (Fig. 1).
2.2 Measurements
For data collection in the NHANES, the information of subjects was collected through interviews, physical examinations, and the collection of blood samples. A self-reported standardized questionnaire was used to collect information about age, race and ethnicity, sex, education, ratio of family income to poverty (PIR) (low: < 1.30, moderate: 1.31 to 3.50; high: ≥ 3.5), and medical history.
The formula of weight in kilograms divided by height in meters squared was applied to calculate the BMI. The trained staff used a mercury sphygmomanometer to measure the blood pressure (BP) of the participants, who had rested for more than 5 minutes. Then, the final value was an average of three measurements (in mm Hg). Responses to inquiries about whether a person had smoked at least 100 cigarettes in their lifetime and if they did so now were used to determine their level of cigarette use. The answers to surveys about whether a participant was drinking now or had at least 12 drinks over their lifetime and had at least 12 drinks in the previous year were used to determine their drinking status. Physical activity (PA) was defined by the questionnaires regarding whether a participant was vigorous or moderate in recreational activities and how much time they usually spent sitting each day.
At the mobile examination center, the blood samples of participants were collected and stored at -20°C before being sent to the central laboratories, where cholesterol, HDL-C, hemoglobin A1c (HbA1c), uric acid (UA), and creatinine (CR) were measured through standard methods. Fasting triglycerides, low-density lipoprotein cholesterol (LDL-C), glucose, alanine transaminase (ALT), and glutamic pyruvic transaminase (AST) were assessed in eight-hour fasting blood samples.
Diabetes was regarded as having been diagnosed with diabetes, being on insulin or taking diabetes pills, having a fasting plasma glucose of ≥ 126 mg/dL (≥ 7.0 mmol/L)8, or having an HbA1c level greater than 6.5%9. Hypertension was defined as having been diagnosed with hypertension or taking antihypertension medication and/or having a high biological measurement value (systolic blood pressure ≥ 130 mm Hg and/or diastolic blood pressure ≥ 80 mm Hg)10. Participants were considered to have dyslipidemia if they were diagnosed with dyslipidemia and were currently taking cholesterol-lowering medications or had a TC ≥ 6.2 µmol/L and LDL-C ≥ 4.1 µmol/L 11. The estimated glomerular filtration rate (eGFR) was calculated by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) Eq. 12.
2.3 Measurement of serum Cu
Inductively coupled plasma dynamic reaction cell mass spectrometry was applied to measure the serum Cu levels of participants in the NHANES. Bulka et al.13. described the detection steps in detail. Notably, the lower detection limit of serum Cu was 0.39 µmol/L. The proportions at or above lower detection limit accounted for 100%.
2.4 Definition of incident CVD
In the NHANES, information about CVDs was collected through a self-reported personal interview. Participants were regarded as having CVDs, including HF, CHD, MI, and stroke, if they responded “yes” to the questions “Has a doctor or other health professional ever told you that you had congestive heart failure?”, “Has a doctor or other health professional ever told you that you had coronary (kor-o-nare-ee) heart disease?”, “Has a doctor or other health professional ever told you that you had a heart attack (also called myocardial infarction)?”, and “Has a doctor or other health professional ever told you that you had a stroke?”
2.5 Covariates
The selected covariates included age, sex, race (Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, and Other Race), marital status (unmarried, married, other), education (primary school graduate or below, middle/high/special school, college graduate or above), BMI, eGFR, albumin, ALT, moderate PA, smoking status, drinking status, and history of the disease (diabetes mellitus, hypertension, dyslipidemia, CKD).
2.6 Statistical analysis
The weighted means (standard error) were used to present the continuous variables, and the weighted frequency percentages were applied to present categorical variables.
Z-standardization (mean = 0, SD = 1) was applied to improve normality and comparability14, and categorical variables (according to tertiles of Cu in the general participants) were used to assess the associations that deviated from linearity.
The population was divided into tertiles based on serum Cu, and the serum Cu tertiles were T1 (< 16.64 µmol/L), T2 (16.64–19.94 µmol/L), and T3 (≥ 19.94 µmol/L). The differences in the three groups' characteristics were compared by employing the chi-squared test or the Kruskal-Wallis H test.
Binary logistic regression analysis was used to evaluate the association between serum Cu and CVD, and the results were expressed as odds ratios (ORs) and 95% confidence intervals (CIs) with three predefined models. The crude mode was unadjusted for covariates; Model I was adjusted for age, sex, and race; Model II was adjusted for Model I, marital status, BMI, education, eGFR, albumin, ALT, diabetes mellitus, hypertension, and dyslipidemia; and Model III was adjusted for Model II, CKD, moderate PA, smoking status, and drinking status. The nonlinear relationships between Cu and CVD were analyzed using restricted cubic spline curves15.
Stratified analyses were performed in various subgroups, including sex, age (< 60 vs. ≥ 60 years old), BMI (< 30 vs. ≥ 30 kg/m2), eGFR (< 90 vs. ≥ 90 mL/min/1.73 m2), hypertension (yes vs. no), diabetes mellitus (yes vs. no), dyslipidemia (yes vs. no), smoking status (never vs. former vs. current), drinking status (never vs. former vs. current), and moderate PA (yes vs. no), in the subgroup analysis compared with the highest tertile and the lowest tertile.
To prevent the missing data from leading to bias, we conducted the multiple-imputation analysis based on 5 replications, and the Markov-chain Monte Carlo method16was used to account for these covariates with missing data, including BMI, eGFR, and drinking status. We also performed sensitivity analyses using a complete-case analysis.
All data analysis was performed using R software, version 4.1.3 (www.R-project.org), and a two-sided P value of < 0.05 was considered statistically significant in all analyses.