Individually-matched etiologic studies: classical estimators made new again.

Area of Science:

• Epidemiology
• Biostatistics
• Medical Research Methodology

Background:

• Individual matching in etiologic studies, once analyzed by classical methods, declined due to cost and analysis complexity.
• The advent of Big Data makes individual matching an economically viable design option.
• Regression-based methods, like conditional logistic regression, are common but often perceived as 'black boxes'.

Purpose of the Study:

• To re-introduce and update classical estimators for individually matched etiologic studies.
• To highlight the flexibility and ease of use of these classical methods.
• To connect classical and regression-based analytical approaches and clarify parameter estimation.

Main Methods:

• Revisiting and updating classical estimators for individually matched case-control studies.
• Demonstrating Maximum Likelihood estimation using summary data for any matching ratio (M:1).
• Two computational approaches are presented: using binomial regression with offsets and manual calculation.

Main Results:

• Classical estimators are flexible, easy to use, and computationally accessible, even with modern datasets.
• Maximum Likelihood estimates can be obtained efficiently using summary data from matched studies.
• The study emphasizes the connection between traditional and modern statistical techniques.

Conclusions:

• Classical methods for individually matched etiologic studies remain relevant and practical.
• Updated classical approaches offer accessible and transparent alternatives for data analysis.
• Understanding the parameter being estimated is crucial for accurate etiologic research.