The effects of recall sequence/mode and day of week on the estimation of usual intake from 24‐hour recall data: An analysis of the 2003–2004 National Health and Nutrition Examination Survey

Abstract

The concept of usual intake is salient to monitoring the nutritional status of populations and to assessing relationships between diet and disease. In estimating means and distributions of usual intake using 24‐hour recall data, it may be important to consider factors such as recall sequence, mode of data collection, and day of week. Using data from the 2003–2004 National Health and Nutrition Examination Survey (n = 7647), we examined the influence of recall sequence and mode (first recall, completed in person versus second recall, completed by phone) and day of week (weekday versus weekend) on mean intakes of energy, food groups, sources of empty calories, and a number of nutrients of public health significance. Although the pattern of effects differed by dietary component and among age/sex subgroups, both sequence/mode and day of week had significant effects on mean intakes. In some cases, the differences appear large enough to be of practical significance. For example, 19–70 year olds reported 100 fewer calories and 2.5 fewer teaspoons of added sugars on average on the second recall compared to the first. Sequence/mode and day of week effects also appear large enough to impact estimated usual intake distributions. The findings highlight the need to consider potential nuisance effects when modeling usual intake.