Bayesian Sequential Meta-Analysis: R Code

Date

Authors

Prashad, Michael
Pearce, Sydney

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

This document contains R coding for Bayesian sequential meta-analysis methods. This statistical methodology allows one to retrospectively create effect size estimates at each point a new study is added to the literature in temporal order or another meaningful organizational order. For example, if 10 studies are included in the analysis, 10 estimates would be generated, starting with an analysis of the first study published, then the first two published studies, then the first three published studies, and onward until all studies have been added. The tenth and final estimate would result in a similar summary effect size as a traditional meta-analysis of all studies. Stopping rules, which are included in the code, can also be used to determine if stopping the analysis due to intervention/parameter effect or stopping due futility in finding effect is appropriate, assessed at each interim analysis. This code was originally developed for a meta-analysis of internal teat sealant efficacy for preventing intramammary infections and clinical mastitis in dairy cows and can be used as an example of how to apply this code. Please reach out to the authors regarding this application (in preparation for journal submission at time of Atrium publication) or if you have questions regarding the use or interpretation of the code.

Description

Keywords

Sequential Meta-Analysis, Bayesian, R Coding, Statistical Methods, Stopping Rules

Citation

Collections