A method for sensitivity analysis to assess the effects of measurement error in multiple exposure variables using...

Area of Science:

• Epidemiology
• Nutritional Science
• Biostatistics

Background:

• Measurement error in self-reported dietary intake can bias associations between diet and health outcomes.
• Mismeasured confounders further distort these associations, leading to invalid conclusions.
• Adjusting for bias is challenging without internal validation data.

Purpose of the Study:

• To propose and evaluate a method for adjusting diet-disease association bias caused by measurement error in dietary intake and a mismeasured confounder.
• To address bias in the absence of internal validation data by incorporating prior information on instrument validity.
• To compare the proposed method against approaches that ignore confounder effects or measurement errors.

Main Methods:

• Developed a statistical method combining prior validity information with observed data to adjust for measurement error bias.
• Conducted sensitivity analyses on varying measurement error magnitudes, error correlations, and validation data uncertainties.
• Applied the method to data on fruit and vegetable intake, smoking, and all-cause mortality from the EPIC study.

Main Results:

• The proposed method increased the strength of the association between fruit and vegetable intake and mortality by approximately fourfold.
• Measurement error in the confounder had minimal impact on the hazard ratio for fruit and vegetable intake with weakly correlated errors.
• The impact of confounder measurement error was more substantial with strongly correlated errors.

Conclusions:

• The developed method allows for sensitivity analysis of measurement error structures and accounts for uncertainties in reported validity coefficients.
• This approach is valuable for assessing the direction and magnitude of bias in diet-disease associations attributable to confounder measurement errors.
• The findings highlight the importance of accounting for measurement error in both dietary intake and confounders for accurate epidemiological research.