Exact inference for the random-effect model for meta-analyses with rare events.

Area of Science:

• Biostatistics
• Clinical Trial Methodology
• Pharmacovigilance

Background:

• Meta-analysis synthesizes data from multiple studies for improved statistical inference.
• Random-effect meta-analysis is used for evaluating treatment safety via adverse event incidence rates.
• Current methods often rely on asymptotic approximations, which are unreliable with limited clinical trials.

Purpose of the Study:

• To develop a robust statistical method for meta-analysis of rare adverse events.
• To address limitations of existing meta-analysis techniques with small sample sizes and rare events.
• To construct an exact confidence interval for the beta-binomial model's location parameter.

Main Methods:

• Developed a novel method for constructing an exact confidence interval.
• Utilized inversion of exact tests for the beta-binomial model.
• Focused on scenarios with a limited number of studies and rare events.

Main Results:

• The proposed confidence interval guarantees coverage at or above the nominal level.
• The method is valid irrespective of the number of studies or within-study sample size.
• Demonstrated applicability to the analysis of rare-event data in clinical trials.

Conclusions:

• The presented method provides a reliable approach for safety assessment in clinical trials with rare adverse events.
• This exact confidence interval method overcomes limitations of asymptotic approximations in meta-analysis.
• Offers a statistically sound tool for evaluating treatment safety when dealing with sparse data.