Ethnic and socioeconomic differences in variability in nutritional biomarkers

Background: Several studies have reported ethnic, education, and income differentials in concentrations of selected nutritional biomarkers in the US population. Although biomarker measurements are not subject to biased self-reports, biologic variability due to individual characteristics and behaviors related to dietary exposures contributes to within-subject variability and measurement error. Objective: We aimed to establish whether the magnitude of components of variance for nutritional biomarkers also differs in these high-risk groups. Design: We used data from 2 replicate measurements of serum concentrations of vitamins A, C, D, and E; folate; carotenoids; ferritin; and selenium in the third National Health and Nutrition Examination Survey second examination subsample (n = 948) to examine the within-subject and between-subject components of variance. We used multivariate regression methods with log-transformed analyte concentrations as outcomes to estimate the ratios of the within-subject to between-subject components of variance by categories of ethnicity, income, and education. Results: In non-Hispanic blacks, the within-subject to between-subject variance ratio for [beta]-cryptoxanthin concentration was higher (0.23; 95% CI: 0.17, 0.29) relative to non-Hispanic whites (0.13; 0.11, 0.16) and Mexican Americans (0.11; 0.07, 0.14), and the lutein + zeaxanthin ratio was higher (0.29; 0.21, 0.38) relative to Mexican Americans (0.15; 0.10, 0.19). Higher income was associated with larger within-subject to between-subject variance ratios for serum vitamin C and red blood cell folate concentrations but smaller ratios for serum vitamin A. Overall, there were few consistent up- or down-trends in the direction of covariate-adjusted variability by ethnicity, income, or education. Conclusion: Population groups at high risk of adverse nutritional profiles did not have larger variance ratios for most of the examined biomarkers.