Study participants
The National Center for Health Statistics ethics review board approved the NHANES protocols and written informed consent was obtained from all participants.
This analysis of deidentified data did not involve direct interaction with participants and was not subject to institutional review board review. NHANES provides data from a representative sample of noninstitutionalized US population.
From 2013 to 2016, a total of 13 104 participants had total estradiol, total testosterone, and sex hormone-binding globulin data.
The following 3 632 participants were sequentially excluded: those without albumin (N = 1 697) or uric acid (N = 3) or creatinine (N = 1) or body mass index data (N = 115), those pregnant (N = 57), those undergoing hysterectomy or ovary removal (N = 966), those taking sex hormone medication including testosterone, progesterone, estrogen, or unspecified sex hormones (N = 170), and those with kidney disease (estimated glomerular filtration rate < 60 mL/min per 1.73 m
2, N = 623). Albumin data were needed to calculate free and bioavailable estradiol or testosterone concentrations
30 and creatinine data were needed to calculate estimated glomerular filtration rate
31. The resulting 9 472 subjects were included in the final analysis of the current study.
Free and bioavailable estradiol and testosterone
Estradiol and testosterone mainly exist in three forms: free, albumin bound, or sex hormone-binding globulin bound. The fraction that is bound to sex hormone-binding globulin is biologically inactive. Bioavailable estradiol or testosterone is the biologically active fraction of estradiol or testosterone, and it includes both free and albumin-bound forms of each hormone. Free and bioavailable estradiol and testosterone were calculated according to the method provided by Vermeulen and colleagues30 based on serum concentrations of total estradiol or testosterone, sex hormone-binding globulin, and albumin.
eGFR calculation and definition of kidney disease
eGFR was calculated by the CKD-EPI Eq. 31: eGFR = 141 X min(Scr/κ, 1)α X max(Scr/κ, 1)−1.209 X 0.993Age X 1.018 [if female] X 1.159 [if black], where Scr is serum creatinine, κ is 0.7 for females and 0.9 for males, α is -0.329 for females and − 0.411 for males, min indicates the minimum of Scr/κ or 1, and max indicates the maximum of Scr/κ or 1. Kidney disease32 was defined as eGFR < 60 mL/min per 1.73 m2 and people with kidney disease were excluded from this study because impaired kidney function significantly increases uric acid levels in the circulation6,10.
Covariates
Confounding covariates included age, ethnicity (Hispanic, non-Hispanic white, non-Hispanic black, or other), body mass index (continuous variable), eGFR (continuous variable, an indicator of kidney function), and self-reported health status (excellent, very good, good, fair, poor, or unknown). Lifestyle confounders included smoking status (never, former, current, or unknown), alcohol consumption (non-drinker, former drinker, current drinker, or unknown), and leisure-time physical activity (0, 1-149, 150–299, and ≥ 300 minutes per week, or unknown)33,34. Leisure-time physical activity was calculated using the following formula34: leisure-time physical activity (min/week) = 2 X vigorous activity frequency X vigorous activity duration + moderate activity frequency X moderate activity duration. Self-reported comorbidities included self-reported diagnoses of hypertension (yes, no, or unknown), diabetes (yes, no, borderline, or unknown), hypercholesterolemia (yes, no, or unknown), coronary heart disease (yes, no, or unknown), stroke (yes, no, or unknown), sleep disorder (yes, no, or unknown), gout (yes, no, or unknown), and cancer (yes, no, or unknown).
Statistical Analysis
Data from the two NHANES cycles (2013-14 and 2015-16) were combined using the appropriate weighting methods35. Four-year weights were calculated by dividing the two-year weights by two36 and used in all analyses to adjust for unequal selection probability and non-response bias following NHANES analytical guidelines35. Population means, medians, and proportions were estimated and reported. Uric acid, age, and eGFR were approximately normally distributed. Estradiol, testosterone, sex hormone-binding globulin, and body mass index were not normally distributed, and these variables were log-transformed before regression analyses were conducted. Descriptive statistics were presented as median and interquartile range (non-normally distributed continuous data), or mean and SD (approximately normally distributed continuous data), or percentages (categorical data). The association analyses were conducted using weighted least squares regression37 with or without adjusting for covariates. Sub-analyses were conducted in the following sub-cohorts: participants aged 12–19 years or 20–80 years or women around menopausal age (47–56 years). All tests were two-sided and a P value of < 0.05 was regarded as statistically significant. All statistical analyses were performed using SPSS version 27.0 (IBM SPSS Statistics for Windows, Armonk, NY, International Business Machines Corporation).