Hazard ratio estimation for biomarker-calibrated dietary exposures

Area of Science:

• Nutritional Epidemiology
• Biostatistics
• Dietary Assessment

Background:

• Self-reported dietary data often contain measurement errors, impacting the reliability of nutritional epidemiology.
• Objective biomarkers (e.g., urinary nutrient recovery) can assess short-term nutrient intake and calibrate self-reported data.
• Existing measurement error models may not fully capture systematic and subject-specific errors in dietary self-reports.

Purpose of the Study:

• To develop and evaluate statistical methods for addressing complex measurement error structures in self-reported dietary data.
• To improve the accuracy of nutritional epidemiology by calibrating self-reported intake with objective biomarkers.
• To extend existing hazard ratio estimation procedures for use with enhanced measurement error models.

Main Methods:

• Utilized a nonstandard measurement error model incorporating systematic, subject-specific, and random error components.
• Extended three estimation procedures for Cox model parameters: risk set regression calibration, conditional score, and nonparametric corrected score.
• Developed an estimator for the cumulative baseline hazard function.

Main Results:

• Assessed the performance of the proposed statistical methods through a comprehensive simulation study.
• Demonstrated the applicability of these methods to real-world data, using an example from the Women's Health Initiative Dietary Modification Trial.
• Provided a framework for more reliable analysis of self-reported dietary data in epidemiological studies.

Conclusions:

• The developed statistical methods effectively address complex measurement error structures in self-reported dietary data.
• These advanced techniques enhance the reliability of nutritional epidemiology by improving the calibration of self-reported intake.
• The findings offer valuable tools for researchers seeking to minimize bias and strengthen conclusions in dietary studies.