The Effect of Editing Open-Ended Text Responses on Nutrient and Food Group Estimates from the Automated Self-Administered 24-Hour Dietary Recall (ASA24)☆

Abstract The National Cancer Institute (NCI) Food Reporting and Comparison Study (FORCS) compares nutrient and food group estimates obtained from recalls collected through the web-based Automated Self-Administered 24-hour Recall (ASA24) to those from standard interviewer-administered telephone recalls among 1054 adults. Within the ASA24 software, respondents may choose to enter open-ended text to report “unfound food” and “other, specify.” Respondents use the“unfound food” field to report a food or drink they cannot find; the ASA24 system then asks a series of general questions to better identify the food. Respondents use the “other, specify” field to provide food details, such as brand name or cooking method, that differ from the responses offered by ASA24. When respondents use these two open-ended text fields, the system assigns default food codes. Nutritionists reviewed these codes and assigned new codes when appropriate. Nutrient and My Pyramid Equivalent (MPE) values from default food codes (unedited) andthe researcher assigned food codes (edited)were compared. Of the 716 food codes evaluated, changes were made to 248 foods (1.5% of total foods reported) which affected 194 (19.1%) of all recalls and required over 60 hours of specialized staff time. Although 43% of the 1013 completed recalls included at least one of the two text field responses, the ASA24 systemassigned acceptable default food codes 68% of the time.Editing did not significantly affect the means and ranking for most nutrients and food group MPE values; correlation coefficients for energy and macronutrient values before and after editing were 0.95 or higher, suggesting that this editingmaynot be necessary for most large studies using ASA24.