Comparison of 4 Methods to Assess the Prevalence of Use and Estimates of Nutrient Intakes from Dietary Supplements among US Adults

Abstract

Background: Accurate and reliable methods to assess prevalence of use of and nutrient intakes from dietary supplements (DSs) are critical for research, clinical practice, and public health monitoring. NHANES has been the primary source of DS usage patterns using an in-home inventory with a frequency-based DS and Prescription Medicine Questionnaire (DSMQ), but little is known regarding DS information obtained from 24-h dietary recalls (24HRs).

Methods: The objectives of this analysis were to compare results from 4 different methods for measuring DS use constructed from two data collection instruments (i.e., DSMQ and 24HR) and to determine the most comprehensive method for measuring the prevalence of use and estimating nutrient intakes from DS for selected nutrients. NHANES 2011-2014 data from US adults (aged ≥19 y; n = 11,451) were used to examine the 4 combinations of methods constructed for measuring the prevalence of use of and amount of selected nutrients from DSs (i.e., riboflavin, vitamin D, folate, magnesium, calcium): 1) DSMQ, 2) 24HR day 1, 3) two 24HRs (i.e., mean), and 4) DSMQ or at least one 24HR.

Results: Half of US adults reported DS use on the DSMQ (52%) and on two 24HRs (mean of 49%), as compared with a lower prevalence of DS use when using a single 24HR (43%) and a higher (57%) prevalence when combining the DSMQ with at least one 24HR. Mean nutrient intake estimates were highest using 24HR day 1. Mean supplemental calcium from the DSMQ or at least one 24HR was 372 mg/d, but 464 mg/d on the 24HR only. For vitamin D, the estimated intakes per consumption day were higher on the DSMQ (46 μg) and the DSMQ or at least one 24HR (44 μg) than those on the 24HR day 1 (32 μg) or the mean 24HR (31 μg). Fewer products were also classed as a default or reasonable match on the DSMQ than on the 24HR.

Conclusions: A higher prevalence of use of DSs is obtained using frequency-based methods, whereas higher amounts of nutrients are reported from a 24HR. The home inventory results in greater accuracy for products reported. Collectively, these findings suggest that combining the DSMQ with at least one 24HR (i.e., DSMQ or at least one 24HR) is the most comprehensive method for assessing the prevalence of and estimating usual intake from DSs in US adults.This trial was registered at clinicaltrials.gov as NCT03400436.

Keywords: NHANES; dietary recall; dietary supplement; measurement error; methodology.

Published by Oxford University Press on behalf of the American Society for Nutrition 2019.

Figures

FIGURE 1

Flowchart of the exclusion criteria used to obtain the analytic samples used in this analysis of US adult participants in NHANES 2011–2014. DSMQ, Dietary Supplement and Prescription Medication Questionnaire; 24HR, 24-h dietary recall.

FIGURE 2

Estimated prevalence (%) of DS use among US adults (aged ≥19 y) by method, NHANES 2011–2014. Values are means. The error bars above each bar are indicative of the estimate ± SE. This sample population makes up 73.6% of the weighted sample population reflected in column 1 (i.e., DSMQ) and column 4 (i.e., DSMQ or at least one 24HR). Estimated prevalence of DS use on the DSMQ, 24HR day 1, or 24HR day 2 was constructed based on whether the participant responded “yes” to the question “Have you used or taken any vitamins, minerals or other dietary supplements in the past 30 days?” or “All day yesterday, [day], between midnight and midnight, did [you/SP] take any vitamins, minerals, herbals or other dietary supplements?” respectively. Analytic sample 1 as used in bar 1 (i.e., DSMQ) and bar 4 (i.e., DSMQ or at least one 24HR) includes individuals aged ≥19 y who were not pregnant or lactating with complete information for the DSMQ. NHANES interview weights were used for these analyses, yielding an overall analytic sample of 11,451 US adults. Analytic sample 2 as used in bar 2 (i.e., 24HR day 1) and bar 3 (i.e., 24HR day 1 and day 2 mean) includes individuals aged ≥19 y who were not pregnant or lactating with complete information for the 24HR day 1. NHANES day 1 dietary weights were used for these analyses, yielding an overall analytic sample of 9954 US adults. *Significant difference in estimated prevalence of DS use when compared with the estimated prevalence of use of the referent group (i.e., 24HR mean). #Significant difference in estimated prevalence of DS use when compared with the estimated prevalence of use of the referent group (i.e., DSMQ only). A P value < 0.05 was considered statistically significant. DS, dietary supplement; DSMQ, Dietary Supplement and Prescription Medication Questionnaire; 24HR, 24-h dietary recall.