Generalized Mantel-Haenszel procedures for 2 x J tables

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Methods

Background:

• The Mantel-Haenszel procedure is a standard method for analyzing stratified 2x2 tables.
• Extending this procedure to tables larger than 2x2 is crucial for more complex epidemiological and clinical studies.

Purpose of the Study:

• To review the generalization of the Mantel-Haenszel procedure for 2 x J tables (where J > 2).
• To present generalized Mantel-Haenszel tests and estimators for a common odds ratio.
• To introduce the generalized Breslow-Day test for assessing homogeneity of odds ratios across strata.

Main Methods:

• Review of existing statistical methodologies.
• Description of generalized Mantel-Haenszel tests for stratified categorical data.
• Explanation of estimators for common odds ratios in multi-level strata.
• Presentation of the generalized Breslow-Day test for homogeneity assessment.

Main Results:

• The paper details the mathematical framework for extending the Mantel-Haenszel procedure to 2 x J tables.
• It provides specific methods for calculating point estimates and confidence intervals for common odds ratios.
• The generalized Breslow-Day test is presented as a tool to evaluate the assumption of consistent effect across strata.

Conclusions:

• The generalized Mantel-Haenszel procedure provides a robust framework for analyzing stratified data beyond 2x2 tables.
• These methods are essential for accurate estimation of effect sizes and for testing heterogeneity in epidemiological research.
• The generalized Breslow-Day test aids in identifying situations where a common odds ratio model is appropriate.