Poster Topical Area: Nutritional Epidemiology
Location: Hall D
Poster Board Number: 744
Objectives: Web-based dietary assessment instruments offer distinct advantages over traditional methods including efficiency of administration and data processing. However, the value of combining multiple web-based 24-hour recalls (HR) with a web food-frequency questionnaire (FFQ) to improve the precision of dietary intake estimates has not been investigated. The aim was to compare combinations of instruments in their ability to measure dietary intakes.
Methods: In this cross-sectional multicenter investigation, adult men and women completed three randomly allocated web-based 24HRs and a web-FFQ within 21 days. Usual intake distributions of nutrients and foods from 24HRs were estimated using the National Cancer Institute (NCI) method with weekend, sequence, age and FFQ (if applicable) as covariates. The reference method (Combination A, Table 1) was based on one 24HR in all participants plus a second recall in 40% of randomly selected participants, consistent with national dietary survey methodologies in Canada.
Results: A total of 1025 participants [50.2% female; mean age (± SD) 42.9 ± 13.5 y] completed all questionnaires. Analyses showed no significant difference among the combinations in estimates of mean usual intake for saturated fats (SFA), sodium, fish, fruits and vegetables (Table 1). Moreover, compared with the reference method, using data from three web-24HRs only slightly narrowed the range of the 95% confidence intervals (CI) for estimated intake of SFA (-6.7%), sodium (-3.1%), fish (-6.5%), fruits (-30.2%) and vegetables (-2.0%). Considering data from a web-FFQ in combination with data derived from three web-24HRs did not further modify the 95%CI range for intake of SFA (-0.7%), sodium (0.8%), fish (-1.2%), fruits (+9.7%) and vegetables (-2.7%).
Conclusions: These preliminary analyses suggest that when using the NCI method, data from a single web-based 24HR repeated in only a subsample of individuals provide stable estimates of mean usual intake for selected nutrients and foods in a large cohort of French Canadians. Using web-24HRs on repeated occasions or in combination with data from a web-FFQ has marginal impact on the precision of estimated intake for most nutrient and foods even for seldom consumed foods such as fish.
Quebec, Quebec, Canada