|
Abstract:
|
Iodine is essential for thyroid hormone production. Urine iodine concentration is an indicator of iodine intake. Since the concentration or dilution of urine depends on an individual's hydration status, normalization of random urine specimens is critical. In this study, we compared two standardization techniques both used to normalize random urine specimen concentrations (creatinine (CRE) and specific gravity (SG)). We applied multiple regression models to urinary iodine concentrations in the NHANES 2007-2008 national survey data with age, gender, and race as predictor variables (BASE model). Then we added CRE and SG, separately, into the BASE model. By adding CRE or SG into the BASE model, R-square increased significantly from 0.0330 to 0.3398 and 0.3286, respectively, for the CRE and SG models. The p-values were compared for all predictor variables among these three models, all were statistically significant except gender from the SG model. On the comparison of ? coefficients for three predictor variables among three models, the patterns of values were similar except for gender. Our study shows both CRE and SG are effective methods for correcting urine iodine concentrations.
|
