How Accurate Is Multiple Imputation for Nutrient Intake Estimation? Insights from ASA24 Data.

Area of Science:

• Nutritional Epidemiology
• Dietary Assessment Methods
• Statistical Imputation Techniques

Background:

• Accurate dietary assessment is vital for nutritional epidemiology.
• Automated Self-Administered 24-hour Dietary Assessment Tool (ASA24) data may contain implausible dietary recalls (IDRs).
• The efficacy of multiple imputation (MI) for high-variability dietary data remains uncertain.

Purpose of the Study:

• To evaluate the accuracy of multiple imputation (MI) in estimating nutrient intake.
• To assess MI performance under various missing data percentages (10%, 20%, 40%).
• To analyze MI's reliability for adolescent dietary data.

Main Methods:

• Utilized 24-hour dietary recalls from 743 adolescents (ages 13-18).
• Simulated missing data at 10%, 20%, and 40% deletion rates.
• Applied multiple imputation via chained equations and compared imputed values to actual nutrient intakes.

Main Results:

• Spearman's rho correlations between imputed and actual nutrient intakes were weak (mean ρ ≈ 0.24).
• Accuracy within ±10% of true values was low for most nutrients (<25%).
• Diet quality scores showed slightly better but still limited accuracy (<30%).

Conclusions:

• Multiple imputation demonstrated poor performance in estimating individual nutrient intake among adolescents.
• MI may not be reliable for accurate nutrient estimates despite potentially preserving sample characteristics.
• Further research is needed to enhance data quality and explore superior imputation methods.