Author:
Nyaga Victoria Nyawira,Arbyn Marc
Abstract
Abstract
Background
Despite the widespread interest in meta-analysis of proportions, its rationale, certain theoretical and methodological concepts are poorly understood. The generalized linear models framework is well-established and provides a natural and optimal model for meta-analysis, network meta-analysis, and meta-regression of proportions. Nonetheless, generic methods for meta-analysis of proportions based on the approximation to the normal distribution continue to dominate.
Methods
We developed , a tool with advanced statistical procedures to perform a meta-analysis, network meta-analysis, and meta-regression of binomial proportions in Stata using binomial, logistic and logistic-normal models. First, we explain the rationale and concepts essential in understanding statistical methods for meta-analysis of binomial proportions and describe the models implemented in . We then describe and demonstrate the models in using data from seven published meta-analyses. We also conducted a simulation study to compare the performance of estimators with the existing estimators of the population-averaged proportion in and under a broad range of conditions including, high over-dispersion and small meta-analysis.
Conclusion
is a flexible, robust and user-friendly tool employing a rigorous approach to evidence synthesis of binomial data that makes the most efficient use of all available data and does not require ad-hoc continuity correction or data imputation. We expect its use to yield higher-quality meta-analysis of binomial proportions.
Funder
Horizon 2020 Framework Programme for Research and Innovation of the European Commission
Publisher
Springer Science and Business Media LLC
Subject
Public Health, Environmental and Occupational Health