I'm presenting at the 44th National Nutrient Databank Conference in Washington, DC!
Presentation Objective
Accurate characterization of nutrient exposure from dietary supplements is essential for epidemiological studies. Dietary Ingredient Database (DSID) was developed to provide laboratory analytically supported estimates of ingredient content in DS products and to improve population intake assessment beyond reliance on label declarations alone. Statistical model updates to incorporate DSID predicted mean values (PMs) under alternative weighting schemes can enhance utilities of DSID for different research purposes.
Results
Equally weighted PMs describe the “average product on the shelf”, whereas market-share weighted PMs better approximate exposure in the population (i.e., “the average product consumed”) by emphasizing high-volume products. The 2 weighted models often generate significantly different PMs. Thus, the choice of weighting can change estimated exposure contrasts in epidemiological research, especially when high-volume brands/formulations are systematically different.
Significance
Updated model-based predicted means and uncertainty estimates will be implemented in DSID web-based calculators to strengthen its applicability for nutritional epidemiology.
Funding Sources
NIH ODS and the USDA
(The exact time and date of my allotted speaking time is still being finalized.)
