Estimating days needed for dietary assessment in pregnancy: a modeling study

Area of Science:

• Nutritional Science
• Maternal Health
• Biostatistics

Background:

• Dietary assessment is vital for understanding diet-health relationships.
• Accurate dietary data collection requires balancing intake variability with participant burden.
• Few methods exist to determine optimal dietary data collection duration, especially during pregnancy.

Purpose of the Study:

• To algorithmically determine the necessary number of dietary data collection days for accurately estimating key nutritional characteristics in pregnant individuals.
• To assess energy, macronutrients, macronutrient density, diet quality (HEI 2020), and intake timing.

Main Methods:

• Analysis of dietary records from 147 pregnant individuals (≤28 days each).
• Application of mixed-effects models to estimate days needed for intake correlation (≥0.90) and accuracy (within 20%).
• Bootstrapping used to create a probabilistic framework for practical consequences.

Main Results:

• Within-person variation exceeded between-person variation for all dietary characteristics.
• Macronutrients required the most days (e.g., 17 days for fat), while diet quality and intake timing required fewer (2-6 days).
• Cohort-based estimates underestimated individual accuracy needs, with only 56% meeting HEI requirements.

Conclusions:

• A novel approach for estimating required dietary assessment days is presented, aiding future study design.
• Existing studies may be underpowered due to insufficient data collection duration.
• Cohort-level accuracy estimates may overstate individual-level precision.