Reducing mean squared error in the analysis of stratified epidemiologic studies

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Methods

Background:

• Case-control studies are essential for investigating disease etiology.
• Estimating common odds ratios is crucial for risk assessment.
• Existing methods for stratified studies may lack optimality.

Purpose of the Study:

• To extend Kalish's approximately optimal estimator for pair-matched case-control studies.
• To adapt the estimator for general stratified and unmatched stratified studies.
• To provide a simple, potentially more accurate, estimation method.

Main Methods:

• Leveraging Kalish's (1990) approximately optimal estimation framework.
• Extending the methodology to accommodate general stratified designs.
• Simplifying the estimator's form for unmatched stratified studies.

Main Results:

• The proposed estimator is an extension of Kalish's work.
• The estimator is applicable to general stratified studies.
• For unmatched stratified studies, the estimator is simple and offers a small correction to standard methods.

Conclusions:

• The extended estimator provides an approximately optimal approach for common odds ratio estimation.
• The method offers a practical and potentially more accurate alternative for stratified case-control studies.
• The simplicity of the estimator for unmatched studies enhances its utility.