Allowing for variations in multivitamin supplement composition improves nutrient intake estimates for epidemiologic studies.

Collecting detailed data on dietary supplement use is time-consuming for study participants and investigators, and this is particularly difficult for multivitamin use because of the many different formulations available. Therefore, many studies simply ask about the frequency of multivitamin use and assign default nutrient composition values to obtain nutrient intakes. Multivitamin supplements are important contributors to total nutrient intakes, but it is not known how default values affect the accuracy of intake estimation. In this study, nutrient intakes were calculated from multivitamins consumed by 26,735 multivitamin users who provided detailed information like product name(s) and frequency of use on a mailed questionnaire. We then recalculated the intakes, using 2 different assumptions about the composition of the multivitamin supplements: 1) a single default composition for all products; and 2) four default compositions, 1 for each subtype of multivitamin, i.e., one-a-day with minerals, one-a-day without minerals, B-complex or stress multivitamins, and antioxidant combinations. A total of 1246 different brands of multivitamins were reported and nutrient composition varied widely. Spearman correlation coefficient analyses, using the 4 default nutrient profiles compared with actual nutrient intakes, were >0.5 (P < 0.001) for 12 of 15 nutrients examined. However, correlations using the single default were lower, with only 5 correlations >0.5. Our findings suggest that a questionnaire designed to assess the composition profiles for 4 types of multivitamin products substantially improves the accuracy of nutrient-intake estimates over one that uses a single default nutrient profile for all multivitamin products.