Evaluation of daily eating patterns on overall diet quality using decision tree analyses

Abstract

Background: Preliminary evidence suggests that meal timing is associated with higher quality diets. Less is known about whether types of food consumed during specific eating episodes (i.e., day-level eating patterns) predict diet quality. Objectives: We investigated the association between day-level eating patterns and diet quality. Methods: Decision tree models were built using 24-h dietary recall data from the National Health and Nutrition Examination Survey 2015 and 2017 cycles in a cross-sectional study. Sixteen food groups and 12 eating episodes (e.g., breakfast, lunch) were included as input parameters. Diet quality was scored using the Healthy Eating Index-2020 and categorized as higher or lower quality diets based on the median score. Mean decrease in impurity (MDI) ± standard deviation determined the relative contribution that day-level eating patterns had on diet quality; higher values represented greater contributions. Results: We analyzed 12,597 dietary recalls from 9347 United States adults who were aged 18 y and older with ≥1 complete recall. Meals (breakfast, lunch, dinner) and respective snacking episodes had the greatest variety of dietary groups that contributed to the Healthy Eating Index-2020 score. Any whole-grain intake at breakfast predicted a higher quality diet (MDI = 0.08 ± 0.00), followed by lower solid fat intake (<8.94 g; MDI = 0.07 ± 0.00) and any plant protein intake at dinner (MDI = 0.05 ± 0.00). Conclusions: Day-level eating patterns were associated with diet quality, emphasizing the relevance of both food type and timing in relation to a high-quality diet. Future interventions should investigate the potential impact of targeting food type and timing to improve diet quality.