Estimating distributions of usual total nutrient intake: A comparison of available methods

Distributions of usual nutrient intake are used to monitor a population's nutritional status. It is vital to consider nutrient intakes from all sources, including dietary supplements (DS). This study compared strategies for estimating distributions of total nutrient intake based on data from the National Health and Nutrition Examination Survey (NHANES). In the 2003‐2004 NHANES, two 24‐hour dietary recalls (24HR) were collected; detailed information about type, consumption frequency, and amount of nutrients from DS was collected through an interviewer‐administered questionnaire. A simulation study compared several approaches for combining the 24HR and DS data for calcium intake. Four methods (ISU, NAS, BCP, and NCI) for adjusting 24HR nutrient data were extended to use DS data in three ways by adding: average intake from DS to adjusted 24HR intakes, average intake from DS to each 24HR intake before adjustment, or single‐day DS dosages to none, some, or all of the 24HRs before adjustment. We find that adding average DS intake to adjusted 24HR intakes leads to unbiased estimates for percentiles, but adding DS intake to 24HR intakes before adjusting can lead to substantial bias. These results suggest that complex relationships existing between DS use and habitual diet must not be ignored.