Exposure Measurement Error Correction in Longitudinal Studies With Discrete Outcomes

Area of Science:

• Environmental Epidemiology
• Biostatistics
• Public Health

Background:

• Environmental epidemiologists frequently assess time-varying exposure histories' impact on health.
• Exposure measurements in longitudinal studies often contain errors, complicating accurate effect estimation.
• Existing methods may yield biased results when dealing with mismeasured exposure histories and discrete health outcomes.

Purpose of the Study:

• To develop and evaluate a statistical method for unbiased estimation of exposure history function effects in longitudinal studies with measurement error.
• To address the challenge of time-varying exposure misclassification in discrete outcome studies.
• To improve the accuracy of estimating chronic exposure effects, such as PM2.5 on anxiety disorders.

Main Methods:

• Development of a novel estimation method tailored for main study/validation study designs.
• Exploration of various estimation procedures within the proposed framework.
• Conducting simulation studies to compare the new method against standard analysis.

Main Results:

• The proposed method demonstrated significant bias reduction in finite samples.
• It improved nominal coverage probability compared to standard analysis.
• Simulations confirmed the method's good performance in handling measurement error.

Conclusions:

• The new method provides unbiased estimates for the effects of mismeasured exposure history functions in longitudinal studies.
• Failure to correct for exposure measurement error can lead to underestimation of chronic health risks, e.g., PM2.5's effect on anxiety.
• This approach is valuable for environmental health research involving complex exposure assessments and discrete outcomes.