Comparison of a Web-Based Frequency Questionnaire for Assessment of Beverage Intake with a Validated 7-Day Web-Diary from Danish Teenagers

Aims: To compare beverage intake measured using a web-based quantitative beverage frequency questionnaire (BFQ) with a 7-day estimated beverage diary (WebDAS), and to evaluate the BFQ’s feasibility.
Study Design: Cross-sectional comparison of the BFQ, which contained 37 beverage types including energy drinks and caffeinated beverages, with the WebDAS.
Place and Duration of Study: Sample: Three 9th grade local authority school classes totalling 73 14-16-year-old students from a suburban area in Copenhagen were recruited. The study was carried out between September 2013 and November 2013.
Methodology: First respondents completed the WebDAS at home, and after 2 weeks they completed the BFQ at school. McNemar’s test, the Wilcoxon signed rank test, Spearman’s rank correlation coefficients, a Bland-Altman plot, weighted Kappa statistics and percentages of exact agreement were used to compare the results of the two methods.
Results: 49 (29 boys; 20 girls) had acceptable data from both the WebDAS, and the BFQ. The mean total beverage intake measured by the two methods differed significantly (BFQ: 1566 vs. WebDAS: 1231g/day, P<.01). The Spearman rank correlations were positive (r=.41-75) for all beverages including energy drinks, and significant for most beverages. Significant agreement for the numbers of consumers was observed between methods, except for soft drinks and chocolate. The mean (SD) difference was 335 (769)g/day, primarily due to an intake of water measured with BFQ almost twice as high as that measured with WebDAS. This was reflected in the Bland-Altman plot and in the percentages of exact agreement, which were lower for water than for other beverages (29% vs. 39-46%).
Conclusion: The BFQ gave results comparable to a 7-day beverage diary (WebDAS) in 14-16-year-olds. With a few adjustments, especially with regard to portion sizes and entries for water, we believe the BFQ will be useful in large population-based studies for assessment of beverage intake.