ZIBGLMM: Zero-Inflated Bivariate Generalized Linear Mixed Model for Meta-Analysis with Double-Zero-Event Studies

Area of Science:

• Biostatistics
• Epidemiology
• Medical Research Methodology

Background:

• Meta-analysis of clinical trials often encounters challenges with double-zero-event studies (DZS).
• Existing methods like continuity correction or study omission can introduce bias.
• Standard models may not account for systemic differences in DZS.

Purpose of the Study:

• To introduce a novel zero-inflated bivariate generalized linear mixed model (ZIBGLMM) for meta-analysis.
• To address the specific challenges posed by DZS in effect size estimation.
• To compare the performance of ZIBGLMM against existing meta-analysis techniques.

Main Methods:

• Developed a two-component finite mixture model (ZIBGLMM) incorporating zero-inflation.
• Created both frequentist and Bayesian versions of the ZIBGLMM.
• Evaluated ZIBGLMM performance using simulations and real-world meta-analysis data, comparing risk ratios (RRs).

Main Results:

• ZIBGLMM demonstrated superior performance in estimating true effect sizes compared to the standard bivariate generalized linear mixed model and conventional two-stage meta-analysis excluding DZS.
• The proposed model achieved substantially less bias.
• Comparable coverage probability was observed for ZIBGLMM relative to the other methods.

Conclusions:

• ZIBGLMM offers a more accurate and less biased approach for meta-analysis involving DZS.
• The model effectively handles zero-inflation, improving risk ratio estimation.
• ZIBGLMM provides a robust alternative to conventional methods for handling DZS in meta-analysis.