Abstract 028: Web-based Delivery with Food Portion Images Improves Dietary Assessment Results
Introduction: Dietary assessment is achieved in a variety of ways, from written records and interviews to short questionnaires based on food categories, all with mixed success. Problems faced by the clinician are limited time for conducting assessments, time and resources required to evaluate collected information and lack of resources for supporting the change process. Due to their low cost and low participant burden, Food Frequency Questionnaires (FFQs) are the most practical method of dietary assessment but improvements are needed.
Hypothesis: VioScreen, developed from a National Cancer Institute grant using a graphical method for dietary data collection and generating reports on nutrient intake and food use patterns will reliably assess intake making it suitable for clinical counseling and research. VioScreen’s web-based self-administered dietary habits questionnaire will provide an efficient and thorough assessment of dietary intake data.
The tool: VioScreen is the only web-based, dietary assessment tool that includes 1,200 food images and portion size options. Results are immediately available for analysis by the researcher/clinician; an optional report can be produced for the participant that includes a food pattern analysis, a list of foods and nutrients consumed, and when used by a counselor generates tailored behavioral feedback. VioScreen uses the University of Minnesota’s Nutrition Coordinating Center database to ensure up-to-date food and nutrient information.
Method: VioScreen was evaluated through an inter-method reliability study with 74 subjects conducted at The Ohio State University by comparing a baseline and 3 month FFQ to six 24-hour recalls conducted between the two FFQ’s. Average time for completing the FFQ was 26.7 +/− 10.0 minutes.
Results: The inter-method reliability was higher for VioScreen than for the paper FFQ VioScreen was modeled after and higher than reported for many other paper FFQs used in major epidemiological studies. Of the macronutrients, only alcohol values were similar; for all others VioScreen correlations were substantially higher, being at or above 0.80 for most macronutrients (0.90 for alcohol, 0.84 for saturated fat, 0.82 for fat, and 0.79 for carbohydrate) and 0.67 for protein. Participant evaluations of VioScreen were generally very good to excellent on ease of use and capturing foods usually consumed. All subjects rated the questionnaire as easy to use, 93% rated VioScreen as either great or excellent, 99% would complete VioScreen if asked by a doctor and 95% thought pictures helped in selecting portion sizes.
Conclusions: Use of multiple portion size pictures, incorporation of complex skip patterns, and elimination of missing or non-interpretable responses contributed to higher inter-method reliability.
- © 2012 by American Heart Association, Inc.