Dietary intake measured from a self-administered, online 24-hour recall system compared with 4-day diet records in an adult US population.

The objective of this study was to compare nutrient intake of two 24-hour recalls collected using the Automated Self-Administered 24-Hour Dietary Recall to a 4-day food record. A convenience sample of university-affiliated adults was chosen because of the diverse population at this university. Ninety-three participants completed the 4-day record and were then prompted to complete two 24-hour recalls within 2 weeks after. Pearson correlation coefficients were calculated for nutrient intake and Healthy Eating Index 2005 (HEI-2005), a summary measure of diet quality. Nutrients and HEI-2005 were also divided into quartiles and percent agreement and κ values were calculated. Results indicated that mean nutrient intakes were similar across the recall and record. Pearson correlations comparing the record and recall ranged from 0.16 to 0.78; with most correlations being between 0.4 and 0.6. For quartiles of dietary intake, percent agreement was moderately high (62.6% to 79.8%), with low to moderate κ values (κ=0.11 to 0.52). The 24-hour recall provided a good overall ranking of intake compared to a 4-day food record. Overall correlations and percent agreement were moderate across the nutrients and HEI-2005, suggesting that the 24-recalls may have been capturing different information than the food record in our population. Individual researchers will need to weigh the benefits of a more automated system, such as efficiency, against the potential loss of food item detail and potential need for larger sample sizes, for their particular study populations.