Validation of the Remote Food Photography Method to Quantify Intake of Infant Formula

Abstract

Childhood obesity rates have more than tripled since the 1970s, and this increased prevalence is cause for concern as childhood obesity increases the risk of adult obesity and other comorbid diseases. Evidence suggests that the origins of obesity can be identified in infanthood. Accurate methods of assessing food intake in infants can be utilized to establish effective feeding practices in infanthood and to assess the relationship between infant feeding practices and the risk of childhood obesity. Current methods are either subjective or have limited ability for widespread use beyond clinical research settings due to cost and high burden. The aim of the Baby Bottle study was to assess the accuracy of the Remote Food Photography Method (RFPM), a novel food intake assessment method, in estimating infant formula as compared to the gold standard, the directly weighed foods method. In the Baby Bottle study, fifty-three adults were recruited to prepare infant formula bottles and use the RFPM to capture photographs of infant formula at different stages of bottle preparation. Dry food provision, liquid food provision, and liquid waste gram weights measured by the RFPM and directly weighed foods method were compared to assess the accuracy of the RFPM in the estimation of infant formula. Paired dependent t-tests and the Bland-Altman regression method were employed to determine if the weight estimations of RFPM differed from the weights measured by the directly weighed foods method. Multivariate analysis of variance was used to analyze the effects of trial number and caregiver status on infant formula preparation. The RFPM estimated liquid formula intake within 10% of the directly weighed foods method, with error of -4.1 ± 14.4% (P<0.0001), 2.8 ± 16.3% (P=0.1550), and 7.0 ± 12.4% (P<0.0001) in 2 fluid ounce, 4 fluid ounce, and 6 fluid ounce bottles, respectively. The RFPM overestimated liquid formula intake by 14.0 ± 10.3% (P<0.0001) in 8 fluid ounce bottles. There were no significant differences between individuals in the caregiver group (n=28) and the non-caregiver group (n=25) based on all demographic and descriptive characteristics. There were no significant differences for the effects of trial number and caregiver status on infant formula preparation except for a significant main effect of caregiver status on the preparation of dry food provision of 2 fluid ounce bottles (P=0.0499) and a significant interaction between trial number and caregiver status on preparation of dry food provision of 4 fluid ounce bottles (P=0.0146). In conclusion, the