JMIR Publications

ABSTRACT

Background:

A main requirement in epidemiologic investigations is the adequate classification of exposure. Web-based questionnaires allow collecting data quickly and with minimal costs from large sample groups through online-self-administered forms. Until now there has been a lack of evidence from large-scale epidemiological studies and nutrition surveys that have evaluated the comparison between traditional and new technologies to measure dietary intake.

Objective:

Our aim was to compare results from the general baseline questionnaire (Q_0) and the 10-year follow up questionnaire (Q_10) in the SUN prospective cohort obtained from different subjects, some of which used a paper-based version and others a web- based version. Both baseline and 10-year assessments included a validated 137-item semi-quantitative food-frequency questionnaire (FFQ) used to collect dietary intake.

Methods:

: The SUN project is a prospective cohort study (with continuous open recruitment and many participants just recently recruited). All participants are university graduates. Participants who completed the validated food frequency questionnaire at baseline (FFQ_0, n=22564) were selected. The variables analyzed were classified into six groups of questions: food-frequency questionnaire (137 items), b) healthy eating attitudes (10 items), c) alcohol consumption (3 items), d) physical activity during leisure-time (17 items), e) other activities (24 items), and f) personality traits (3 items). Multiple linear and logistic regression models were used to assess the adjusted differences between the mean number of missing values and the risk of having apparently incorrect values for FFQ items or mismatches/inconsistencies in dietary variables.

Results:

Only 339 (1.5 %) and 6765 (60.7 %) participants reported their information using the web-based version for Q_0 and Q_10, respectively and 51.4 % and 100 % of participants who completed the Q_0 and Q_10 respectively, had the option of choosing the web-based version. Socio-demographic, lifestyle, health-characteristics, food consumption, and energy and nutrient intakes were similar between participants according to the type of questionnaire used in Q_10. Less than 0.5 % of values were missing for items related to healthy eating attitudes, alcohol consumption and personality traits in the web-based questionnaires. The proportion of missing data in FFQ, leisure-time physical activity and other activities was higher in paper-based questionnaires than in their web-based counterparts. In web-based questionnaires, a high degree of internal consistency was found when comparing answers that should not be contradictory, such as the frequency of fruit as dessert vs total fruit consumption and the frequency of fried food consumptions vs oil consumption.

Conclusions:

The incorporation of a web-based version for a baseline and 10-year questionnaire has not implicated a loss of data quality in this cohort of highly educated adults. The preference for web-based questionnaires was greater in younger participants. Web-based questionnaires were filled out to a greater extent and with less missing items than paper-based questionnaires. Further research is needed to optimize data collection and response rate in web-based questionnaires.