Poster Abstracts

Background: The Remote Food Photography Method (RFPM) accurately estimates energy intake in adults; however, its validity in adolescents is not yet supported. Adolescents often exhibit the lowest accuracy in self-reported energy intake, with adolescents living with obesity having larger errors than those with healthy weight. This free-living study examined the validity of the RFPM and SmartIntake app in estimating energy intake over 3 days compared to estimated intake using gold standard, Doubly Labeled Water (DLW) among adolescents. Methods: Adolescents were dosed with DLW and recorded their energy intake for 3 consecutive days using the SmartIntake app. The app was used to capture images of both food selection and plate waste, which were then uploaded to a server for energy intake estimation. Ecological Momentary Assessment (EMA) was used to promote data quality and completeness. The gold standard measure for energy intake was calculated using DLW-assessed energy expenditure, including 3 days that overlapped with the food intake assessment. Equivalence was evaluated with two one-sided t-tests (±10% bounds), and Bland-Altman analysis determined if error variance differed over levels of intake. Mean percent error (MPE) was calculated at the group level and linear regressions were performed to test whether the RFPM's error varied over levels of body mass index (BMI) percentile. Results: Among the enrolled adolescents (n = 22), 17 (77%) were female, 12 (55%) were Black, 5 (23%) had overweight or obesity, and 6 (27%) had severe obesity. The mean (SD) age was 14.3 (1.8) y and BMI percentile was 75.9 (29.7). Mean (SD) energy intake estimated with the RFPM (1618.7 [685.9] kcal/day) was not equivalent (p = 0.99) to mean estimated intake with DLW (2401.5 [623.1] kcal/day). The mean energy intake difference (SD) between the two methods was 782.8 (947.0) kcal, and the RFPM's MPE for estimated energy intake was 32.6%. The Bland-Altman plot for energy intake indicated that the measurement error of RFPM was consistent over different levels of energy intake (p = 0.67). Specifically, participants' errors in estimating energy intake were similar at lower and higher levels of energy intake. Four adolescents with severe obesity failed to comply with the data collection procedures, did not respond to EMA prompts, and did not provide complete records of their eating occasions. When these adolescents were excluded, the MPE was reduced to 20.5%. Regression analysis also indicated that higher BMI percentile was associated with greater underestimates in energy intake (R2 = 0.19; p = 0.04). Conclusions: In free-living conditions, the RFPM and SmartIntake did not accurately estimate energy intake when compared to gold standard, DLW. Adolescents with severe obesity showed higher error rates and were less likely to comply with study procedures. These findings highlight the need for better and more accurate dietary assessment methods in adolescents, especially those with severe obesity.