(P15-011-23) Investigating Eating Architecture: How Precise Does Time of Eating Have To Be?

Objectives: Eating architecture (EA) refers to the size, timing and frequency of eating. Dietary assessment tools vary in how precisely eating occasions (EOs) are distinguished from each other. We characterise EA based on precise and broad EOs definitions and compare associations with covariates. High-quality evidence on the implications of EA is imperative for optimal health guidelines development.

Methods: Data came from 3-day diet diaries for n=4855 children aged 7y in the Avon Longitudinal Study of Parents and Children who had meal slot times (broad; "BD") and exact times (precise; "PD") available. We computed eating size, timing and frequency variables for both EO definitions, calculated intraclass correlation coefficients (ICC), and Bland-Altman analysis estimated the mean difference and limits of agreement (LOA) between them. Secondary analysis included correlations of EA with total energy intake (TEI), waist circumference, body mass index, obesogenic dietary pattern score and sleep duration.

Results: BD vs. PD led to underestimation of frequency (2.2 times/day; LOA -1, 5) and overestimation of EO size (83 g; LOA -179, 13), last time (50 min; LOA -142, 42), and total period (49 min; LOA -161, 63) of eating. Overestimation of size was uniform, while BD and frequency disagreed more from PD at higher frequencies, later initiation and earlier cessation of eating, and shorter eating windows. Larger EO size was associated with a lower frequency, accentuated for PD (r_PD=-0.54 (95% CI -0.56, -0.52) vs. r_BD=-0.23 (-0.25, -0.20)). Whereas the association of higher frequencies with later timing; and larger EO size with high TEI, was weaker for PD (r_PD=0.48 (0.46, 0.50) vs. r_BD=0.65 (0.64, 0.67) and (r_PD=0.27 (0.24, 0.29) vs. r_BD =0.45 (0.43, 0.48) respectively). Consistency of agreement was good to excellent for size (ICC=0.75) and timing (0.58 to 0.80); and moderate for frequency (0.43).

Conclusions: Precise times improve the description of EA, while broader times underestimate eating frequency, overestimate EO size and eating period. However, associations were generally consistent in magnitude and direction by timing precision. Therefore, using less precise times could be an efficient and effective approach for investigations in nutritional epidemiology.

Funding Sources: ANID (Government Support – Non-U.S.); University of Bristol; Wellcome Trust and MRC.