Development of a Live Coding Scheme to Assess Food Acceptance Among Infants and Toddlers in Naturalistic Settings

Abstract Objectives Small-quantity lipid-based nutrient supplements (SQ-LNS) are designed to address malnutrition among infants and toddlers during complementary feeding. There is currently limited research assessing children's acceptability (i.e., behavioral response to the taste) of these supplements, and acceptability may predict long term consumption of these foods. We aimed to adapt a video coding protocol for use in naturalistic settings to determine child food acceptance. Methods Infants (ages 7 to 24 months) were fed two versions of an SQ-LNS (sweetened, unsweetened) mixed with a familiar food (e.g., infant oatmeal, banana) by mothers at two different study sites: a laboratory in Denver, Colorado (n = 56; 86% Non-Hispanic White) and family homes in the state of Morelos, Mexico (n = 62). Trained coders used a video coding (VC) scheme to rate child acceptance of each spoon offer using a 4-point scale (0 = refusal, 1 = enforced, 2 = acceptance, 3 = anticipation). The VC scheme was subsequently adapted for use as: 1) a live coding (LC) scheme to be used in participant homes, and 2) a video live coding (VLC) scheme to be used as a reliability check on the LC scheme. The LC scheme was tested with a sample of infants and caregivers (n = 20) selected using stratified sampling [infant age, supplement group (sweetened, unsweetened)] from the Mexico trial. One coder rated the infants’ responses to the SQ-LNS while observing the feeding interaction in the home. A second coder rated the same infants’ responses while later watching a video of the feeding (VLC). To simulate LC, the second coder did not pause the video. Intra-class correlation coefficients (ICC) were calculated to assess inter-rater reliability between sites and coding methods. Results Reliabilities for VC between Denver and Mexico coders were excellent (ICC > .94), as were reliabilities for coders within study sites (ICCs > .95). Reliabilities calculated between coders utilizing different coding methods were moderate (VLC and LC: ICC = .75), good (LC and VC: ICC = .87), or excellent (VC and VLC: ICC = .93). Conclusions Our LC scheme demonstrated good reliability with a VC scheme to assess infant acceptability of an SQ-LNS. Since traditional VC is both resource and time-intensive, our LC scheme may be useful for assessing infant food acceptance in resource-limited settings. Funding Sources Global Alliance for Improved Nutrition (GAIN).