Adaptation of the Remote Food Photography Method to Assess Infant Intake During Bottle-Feeding of Ready-to-Feed Formula

Abstract

Bottle-fed infants are at a higher risk for obesity, yet interventions to modify bottle-feeding patterns have shown limited success. Accurate assessment of bottle-feeding patterns is an important basis for targeted interventions. Caregiver reports are susceptible to bias, highlighting the need for more robust methods. One promising approach is the Remote Food Photography Method (RFPM), which involves analysis of digital photographs to objectively assess intake. The purpose of this study was to adapt RFPM to measure the amount of ready-to-feed formula in bottles. Research assistants took digital photographs of bottles with varying amounts of formula and assessed bottle weight to create a predictive model data set (n = 100) and an external validation data set (n = 100). Image processing software was used to determine the area of formula in pixels. Formula area was regressed on actual bottle weight and the resulting regression parameters were used to calculate estimated bottle weights. Predictive models were validated by applying them to the external validation data set. Within the predictive model data set, the mean difference between estimated and actual bottle weight was equivalent within ±10% equivalence bounds (0.0005 g [90% CI, −0.45, 0.45]). Within the external validation data set, the mean difference between estimated and actual bottle weight was also equivalent within ±10% equivalence bounds (−17.0 g [90% CI, −18.85, −15.17]). Findings were similar when bottle weight was converted to calories. In sum, the present study provided preliminary evidence for the feasibility and accuracy of an adapted RFPM to assess changes in the amount and caloric content of ready-to-feed formula in bottles.