Study design and population
National Health and Nutrition Examination Survey (NHANES), conducted by the Centers for Disease Control and Prevention (CDC), was an ongoing repeated cross-sectional study designed to assess the lifestyle, health, and nutrition status of non-institutionalized civilian US population through complex, multistage probability sampling [21, 22]. NHANES study protocols were approved by the research ethics review board of the National Center for Health Statistics. Written informed consent was acquired from each participant and obtained from the guardians of participants younger than 18 years of age, and assent was obtained from those aged 12 to 17 years. More detailed information is available at http://www.cdc.gov/nchs/nhanes/nhanes_questionnairees.htm. The NHANES datasets are available on DataDryad (https://doi.org/10.5061/dryad.d5h62).
This study used data from the 1999 to 2006 NHANES. Fasting blood samples were collected from those 12 years of age and older. In total, 9493 eligible adolescents aged 12–19 years were enrolled. We excluded participants with missing SUA value (n = 1184) and BLL value (n = 6).
Exposure variable and outcomes
In our study, the exposure variable was BLL. BLL was measured at the CDC’s National Center for Environmental Health, Division of Laboratory Sciences using a Perkin-Elmer model SIMAA 6000 simultaneous multielement atomic absorption spectrometer with Zeeman background correction. The analytical laboratory followed extensive quality control procedures. National Institute of Standards and Technology Standard Reference Materials whole-blood materials were used for external calibration. The limit of detection (LOD) for blood lead was 0.3 μg/dL in NHANES from 1999 to 2004. The LOD for blood lead was 0.25 μg/dL in NHANES from 2003 to 2006. Lead concentration below the level of detection was assigned the limit of detection divided by the square root of 2, as recommended by NHANES. The interassay coefficients of variation ranged from 3.1 to 4.0% for BLL. Details of these assays and quality standards are available at http://cdc.gov/nchs/nhanes.
The outcome variable was SUA. SUA was measured on a Roche Hitachi Model 917 or 704 Multichannel Analyzer in 1999–2001 and a Beckman Synchron LX20 in 2002 using a colorimetric method. Serum creatinine level was measured with the same instruments using the Jaffe kinetic alkaline picrate method. Serum creatinine was corrected to standardize to a “gold” standard reference method as recommended by NHANES. The distribution of creatinine and uric acid results from the two laboratories were compared at the time of transition, and no significant differences were observed.
Estimated glomerular filtration rate (eGFR, milliliters per minute per 1.73 m2) was calculated via the creatinine-based formula of Schwartz: eGFR = k (height in centimeters)/(serum creatinine in milligrams per deciliter), where k is 0.7 in boys and 0.55 in girls [23, 24].
Covariates
We selected the appropriate covariates based on previous studies examining risk factors for SUA as well as adjusting for covariates that, when added to this model, changed the matched odds ratio by at least 10% [25, 26]. Therefore, the following variables were used to construct the fully adjusted model: continuous variables included age (years), body mass index (BMI, kg/m2), blood pressure (BP, mmHg), poverty to income ratio, dietary data [calcium (mg), total monounsaturated fatty acids (TMFA, gm), total polyunsaturated fatty acids (TPFA, gm), total saturated fatty acids (TSFA, gm), total fat (gm), protein (gm)] and laboratory data [blood cadmium (ug/L), serum cotinine (ng/mL), hemoglobin (g/dL), fasting blood glucose (FBG, mg/dL), total cholesterol (TC, mg/dL), triglycerides (mg/dL), high density lipoprotein cholesterol (HDL-C, mg/dL), blood urea nitrogen (BUN, mg/dL), C-reactive protein (CRP, mg/dL)]; categorical variables consisted of race/ethnicity (non-Hispanic White, non-Hispanic Black, Mexican American, other Hispanic or other), education (less than high school, high school/equivalent, or greater than high school), physical activity (sedentary, low, moderate, high).
Statistical analysis
The statistical analyses were performed according to the guidelines of the CDC (https://wwwn.cdc.gov/nchs/nhanes/tutorials/default.aspx). Sample weights were used for analyses to account for the complex survey design and non-response of NHANES. We calculated the sample weight for the 8 years of data from 1999 to 2006 as WT99–06 = (1/2) × WT99–02 + (1/4) × WT03–04 + (1/4) × WT05–06, where WT99–02 is the variable WTMEC4YR from the NHANES 1999–2000 and NHANES 2001–2002; WT03–04 and WT05–06 were the variable WTMEC2YR from the NHANES 2003–2004 and NHANES 2005–2006 demographic file, respectively [27].
The major question of interest was whether BLL would be associated with SUA after the data was controlled for confounders which might affect SUA. We first examined the distribution of BLL according to the population characteristics listed above. Weighted means, proportions and standard error (Se) were calculated for baseline characteristics using survey sample weights. Because the distribution of values for BLL was strongly skewed toward the upper end, the BLL was Ln-transformed for analysis. Second, we applied multivariate linear regression analysis to evaluate the independent association between LnBLL and SUA. We used four levels of adjustment: Model 1 was adjusted for sociodemographic variables; model 2 was further adjusted for blood biochemical examinations; model 3 was additionally adjusted for dietary data. When model 4 was adjusted for sex, age, BMI, race, hemoglobin, HDL-C and eGFR covariates, the matched odds ratio changed by at least 10%. Third, although there is no universally definition of hyperuricemia in adolescents, previous studies reported that a SUA ≥ 5.5 mg/dL was associated with the risk of hypertenison [16, 28]. Based on this association, the elevated SUA was defined as ≥5.5 mg/dL. We investigated the association of LnBLL with elevated SUA using multivariate binary logistic regression analysis.
The subgroup analyses were performed using stratified multivariate regression analyses. To further characterize the shape of the relationship between BLL and SUA, we used a generalized additive model (GAM) and a fitted smoothing curve (penalized spline method). To ensure the robustness of data analysis, we also did the sensitivity analyses. We converted the LnBLL into a categorical variable, and calculated the P for trend. The purpose was to verify the results of LnBLL as the continuous variable and to observe the possibility of nonlinearity.
All the analyses were performed using the statistical package R (http://www.R-project.org, The R Foundation) and Empower (R) (www.empowerstats.com; X&Y Solutions, Inc., Boston, MA). A two-sided P-value < 0.05 was considered statistically significant.