Confidence-interval construction for rate ratio in matched-pair studies with incomplete data

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Statistical Inference

Background:

• Matched-pair designs enhance efficiency in clinical trials for binary outcomes.
• Rate ratio is a key metric for comparing treatment efficacy.
• Incomplete data presents challenges in analyzing matched-pair trial results.

Purpose of the Study:

• To propose novel confidence interval estimators for the rate ratio in incomplete matched-pair designs.
• To evaluate the performance of these new estimators.
• To provide practical tools for analyzing clinical trial data with missing information.

Main Methods:

• Development of 10 confidence-interval estimators for the rate ratio.
• Proposal of a hybrid method to derive variance estimates from single proportion confidence limits.
• Evaluation of estimators based on coverage probability, interval width, and noncoverage probabilities.

Main Results:

• The hybrid Agresti-Coull confidence interval, utilizing Fieller's theorem, demonstrated satisfactory performance.
• This method proved effective for both small and moderate sample sizes.
• The proposed intervals were illustrated using two real-world clinical trial examples.

Conclusions:

• The hybrid Agresti-Coull confidence interval offers a reliable method for rate ratio estimation in incomplete matched-pair designs.
• The closed-form solution of the hybrid method enhances its practical applicability.
• This research contributes valuable statistical tools for clinical trial data analysis.