Optimal allocation of resources in a biomarker setting

Area of Science:

• Nutritional epidemiology
• Biostatistics
• Dietary assessment methodology

Background:

• Dietary assessment instruments like food frequency questionnaires (FFQ) and diet records (DR) are prone to substantial measurement error.
• Correlated errors between FFQ and DR can bias standard regression calibration methods for estimating nutrient intake.
• Biomarkers, if errors are independent, can enable unbiased estimation of true nutrient intake using the method of triads.

Purpose of the Study:

• To optimize the proportion of subjects with replicated biomarker measures in validation studies.
• To minimize the variance of the estimated regression coefficient (λ) of true intake on surrogate intake.
• To determine the most cost-effective design for biomarker validation studies.

Main Methods:

• Utilized the method of triads for unbiased estimation of the regression coefficient (λ).
• Assessed the impact of different proportions of replicated biomarker measures on the variance of ln(λ̂).
• Applied the methodology to vitamin C intake data from the European Prospective Investigation into Cancer and Nutrition (EPIC) study.

Main Results:

• Identified that the method of triads requires replicate biomarker data for unbiased estimation.
• Demonstrated that optimizing the proportion of replicated biomarker measures is crucial for minimizing estimation variance.
• Found that 21% of subjects with replicated biomarker measures is optimal for the vitamin C intake example.

Conclusions:

• The method of triads provides a robust approach to correct for correlated measurement errors in dietary assessment.
• Optimizing biomarker validation study design, specifically the proportion of replicated measures, is essential for precise nutrient intake estimation.
• The findings offer practical guidance for designing future nutritional epidemiology studies to enhance data accuracy and reliability.