Comparing four methods to estimate usual intake distributions.

Area of Science:

• Nutritional epidemiology
• Dietary assessment methodology
• Statistical modeling

Background:

• Estimating usual nutrient and food intake distributions is crucial for nutritional studies.
• Existing methods have limitations, especially when true intake distributions are unknown.
• The European Food Consumption Validation (EFCOVAL) Study provided real-world data for empirical comparison.

Purpose of the Study:

• To compare the performance of four statistical methods for estimating usual dietary intake distributions.
• To evaluate methods using both simulated data and empirical data from the EFCOVAL Study.
• To identify factors influencing the accuracy of these dietary assessment methods.

Main Methods:

• Comparison of the Iowa State University (ISU), National Cancer Institute (NCI), Multiple Source Method (MSM), and Statistical Program for Age-adjusted Dietary Assessment (SPADE) methods.
• Simulation studies with varying sample sizes, variance ratios, and normality assumptions.
• Empirical analysis using 24-hour dietary recalls and food frequency data from the EFCOVAL Project.
• Evaluation based on mean bias and bias variation for estimated means and percentiles.

Main Results:

• All four methods (ISU, NCI, MSM, SPADE) showed small mean bias overall.
• Bias and its variation increased with smaller sample sizes, higher variance ratios, and non-normal distributions.
• The NCI method exhibited significant bias in specific scenarios (e.g., 95th percentile with r(var)=9, λ(BC)=0, n=1000).
• Empirical analysis on EFCOVAL data showed close agreement (<5%) for nutrient estimates, with larger differences for food groups.

Conclusions:

• The four compared methods generally provide good estimates of usual nutrient intake distributions.
• Caution is advised for nutrients with high within-person variation, skewed distributions, or when using small sample sizes.
• Method selection may be influenced by specific features and practical considerations, as each method has unique strengths.