Weighted estimation for confounded binary outcomes subject to misclassification

Area of Science:

• Epidemiology
• Biostatistics
• Health Services Research

Background:

• Causal effect identification relies on the consistency assumption, which can be compromised by outcome misclassification.
• Differential misclassification bias can distort findings in health research, particularly in administrative datasets.
• Existing methods may not adequately address confounding and misclassification simultaneously.

Purpose of the Study:

• To introduce and evaluate an inverse probability weighted (IPW) approach for estimating marginal causal odds ratios.
• To mitigate bias stemming from differential outcome misclassification in the presence of confounding.
• To assess the finite sample properties of the proposed IPW estimators using simulated data.

Main Methods:

• Developed an IPW method to rebalance covariates across treatment groups.
• Utilized internal validation data for estimating the marginal causal odds ratio.
• Extended the approach to incorporate additional covariates and employed logistic regression for weight estimation.
• Applied a bootstrap method for robust variance estimation.

Main Results:

• The proposed IPW approach effectively rebalances covariates, reducing differential misclassification bias.
• The method demonstrates utility in estimating causal odds ratios even with misclassified outcomes.
• Simulated data analysis confirmed the finite sample properties of the weighted estimators.

Conclusions:

• The inverse probability weighted approach offers a robust strategy for causal inference when outcome misclassification is present.
• This method enhances the reliability of findings from observational health studies, particularly those using administrative data.
• The technique provides a valuable tool for epidemiologists and biostatisticians dealing with data quality issues.