(P31-078-23) Validity of the PortionSize and MyFitnessPal Mobile Apps for Estimating Energy Intake by Demographic Characteristics: A Randomized Laboratory-Based Evaluation

Objectives: The PortionSize app (PS) offers real-time feedback on energy (kcal) intake by using embedded food templates of known volume and a subset of the Food and Nutrient Database for Dietary Studies. This study tested the validity of PS for estimating energy intake during a simulated meal in a laboratory setting, compared the validity of PS to an established and widely adopted app, MyFitnessPal (MFP), and assessed if error for PS and MFP differed by demographic characteristics.

Methods: Participants (N=42, 18-70 years, BMI 18.5-50 kg/m²) were randomized to use PS and MFP to estimate energy intake during a laboratory-based meal, which was covertly weighed (criterion method), and completed a demographics survey. Two One-Sided t-tests assessed equivalence (±21% bounds) between energy (kcal) from weighed food and energy estimated with PS and MFP. Differences in error between PS and MFP were evaluated using dependent samples t-tests. Planned regression-based exploratory analyses determined if error from PS and MFP varied by age, BMI, race, or sex.

Results: The sample was 54.8% female, and majority were white (66.7%). Mean age was 36.3 (±16.6) years, and mean BMI was 28.2 (±6.9) kg/m². For PS, energy was underestimated by 13.3%, which was not equivalent to weighed values (p=0.13). For MFP, energy intake was overestimated by 7%, which was equivalent to weighed estimates (p=0.04). The difference in error between PS and MFP was not significant (p=0.75). For PS, energy was under-estimated by 284.9 kcal for whites and over-estimated by 119.6 kcal for non-whites. This difference in error by race for PS was significant (p=0.024), and error did not differ by age, BMI, or sex (p-values >0.05). For MFP, energy was under-estimated by 95.5 kcal for females and over-estimated by 292.0 kcal for males. This difference in error by sex for MFP was significant (p=0.022), and error did not differ by age, BMI, or race (p-values >0.05).

Conclusions: Energy intake for MFP, but not PS, was equivalent to weighed values. MFP performed well, while PS needs improvements to decrease error. Error in energy intake differed by race for PS, and by sex for MFP, which helps inform modifications to PS and highlights the importance of validating dietary assessment apps among a demographically diverse sample.

Funding Sources: R01 DK124558U54, U54 GM104940, P30 DK072476, T32 DK064584, PBRC-2018-LIFT-001