The effect of misclassification in matched-pair case-control studies

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Exposure misclassification is a common issue in epidemiological studies.
• Pair-matched case-control studies are susceptible to bias, especially with correlated exposure variables.

Purpose of the Study:

• To quantify the excess bias caused by nondifferential misclassification in matched-pair case-control studies.
• To investigate how differential misclassification affects bias in matched studies.
• To present a method for correcting misclassification bias in matched-pair designs.

Main Methods:

• Illustrating excess bias in various case-control scenarios.
• Analyzing the impact of matching on bias under differential misclassification.

Main Results:

• Nondifferential misclassification leads to more severe bias and power loss in matched studies compared to unmatched ones.
• The extent of this "excess bias" is directly related to the closeness of matching on exposure correlates.
• Differential misclassification can either increase or decrease bias, depending on the study specifics.

Conclusions:

• Matching on exposure correlates exacerbates bias from nondifferential misclassification in case-control studies.
• A method is proposed to correct for misclassification bias in matched-pair studies, offering a potential solution for researchers.