Recommended confidence intervals for the conditional odds ratio in matched-pairs designs

Area of Science:

• Biostatistics
• Statistical Inference
• Epidemiology

Background:

• Conditional odds ratios are crucial in matched-pair analyses.
• Limited research exists on confidence intervals for these ratios.
• Existing methods for interval estimation can yield varied results.

Purpose of the Study:

• To investigate and compare different methods for interval estimation of the conditional odds ratio.
• To determine the most suitable confidence interval methods for various sample sizes in matched-pair designs.
• To provide practical recommendations for researchers.

Main Methods:

• Derived four confidence intervals: delta test, score test, inferential model test, and fiducial test.
• Employed exact calculation studies to assess method performance.
• Utilized a real-world example for illustration.

Main Results:

• The four methods evaluated produced different confidence interval results.
• Performance varied depending on sample size (small vs. moderate-to-large).
• Specific methods are recommended for different sample size scenarios.

Conclusions:

• The choice of confidence interval method impacts results in matched-pair analysis.
• Recommendations are provided to guide the selection of appropriate interval estimation techniques.
• This study enhances the understanding and application of conditional odds ratio analysis.