Regression modeling plan for 29 biochemical indicators of diet and nutrition measured in NHANES 2003-2006

The collection of articles in this supplement issue provides insight into the association of various covariates with concentrations of biochemical indicators of diet and nutrition (biomarkers), beyond age, race, and sex, using linear regression. We studied 10 specific sociodemographic and lifestyle covariates in combination with 29 biomarkers from NHANES 2003-2006 for persons aged ≥ 20 y. The covariates were organized into 2 sets or "chunks": sociodemographic (age, sex, race-ethnicity, education, and income) and lifestyle (dietary supplement use, smoking, alcohol consumption, BMI, and physical activity) and fit in hierarchical fashion by using each category or set of related variables to determine how covariates, jointly, are related to biomarker concentrations. In contrast to many regression modeling applications, all variables were retained in a full regression model regardless of significance to preserve the interpretation of the statistical properties of β coefficients, P values, and CIs and to keep the interpretation consistent across a set of biomarkers. The variables were preselected before data analysis, and the data analysis plan was designed at the outset to minimize the reporting of false-positive findings by limiting the amount of preliminary hypothesis testing. Although we generally found that demographic differences seen in biomarkers were over- or underestimated when ignoring other key covariates, the demographic differences generally remained significant after adjusting for sociodemographic and lifestyle variables. These articles are intended to provide a foundation to researchers to help them generate hypotheses for future studies or data analyses and/or develop predictive regression models using the wealth of NHANES data.