Comparison of NHANES 2007‐2010 methodologies for assessing dietary supplement use, part I (245.4)

In NHANES 2007‐2010 dietary supplement (DS) intake, traditionally measured using a DS questionnaire (DSQ) that assessed usage over the past 30 days, was also measured by up to two 24‐hour dietary recalls (24HR). This study investigates how choice of DS assessment method affects estimates of total nutrient intake. Among women aged 51‐70 in the 2007‐10 NHANES (n=1740), 65.5±.02% (estimate±s.e.) reported usage of a calcium supplement on the DSQ or on at least one 24HR. Almost 14±1% of all DS reports came from 24HRs alone, and 19±1% came from the DSQ alone. This led to significantly smaller estimated fractions of DS users when only 1 modality was considered: 56.7% for 24HRs and 59.6% for DSQ (p<.001 per test). The intercept and slope of the linear relationship between 24HR relative frequency (RF) of usage and average reported DSQ usage probability were 0.12±0.04 and 0.74±0.04 respectively, suggesting the DSQ under‐reports low usage and over‐reports high usage. Typically, 24HR‐only DS reports appeared on both 24HRs (RF=0.79±0.03), similar to reports with DSQ values of 25‐29/30 days (RF=0.80±0.07). Calibrating DSQ probabilities to 24HR RF led to a 42±14 / 139±36 mg reduction in the mean/85th percentile of usual DS calcium intake, since most (67±1.6%) of DSQ reports were 30/30 days, suggesting that choice of DS assessment may influence estimation of distributions of total nutrient intake for DS users, especially in the upper tails.