A multivariate method for measurement error correction using pairs of concentration biomarkers

Area of Science:

• Behavioral Epidemiology
• Nutritional Epidemiology
• Biomarker Research

Background:

• Measurement error is a significant challenge in behavioral epidemiology, limiting the accuracy of existing correction methods due to their often unrealistic assumptions.
• Accurate estimation of exposure-effect relationships is crucial for understanding health outcomes.

Purpose of the Study:

• To propose a novel multivariate method for correcting measurement error in behavioral epidemiology.
• To develop a method with more tenable assumptions compared to existing approaches.
• To provide corrected estimates for exposure-effect relationships.

Main Methods:

• Utilizes two concentration biomarkers per nutritional variable and structural equation modeling.
• Employs standardized regression calibration to estimate effects of true exposure variables.
• Preserves hypothesis testing in original units.

Main Results:

• Simulated data analysis demonstrated substantial corrections.
• Sensitivity analysis confirmed the advantage of the corrected approach, showing reduced bias.
• Adequate precision requires relatively large calibration studies.

Conclusions:

• Careful selection of concentration biomarkers enables accurate error-corrected multivariate hypothesis testing and standardized effect estimation.
• The proposed method generally yields less biased results than uncorrected analyses, even with modest deviations from assumptions.