Network meta-analysis (i.e. mixed treatment comparison) is becoming more and more important in evidence synthesis. An interesting aspect of this type of meta-analysis is the possibility to explore the impact of moderators/effect modifiers in a complex evidence network. There are already several posts on this topic in CV:
- Main effects and interaction in multivariate meta-analysis (network meta-analysis) in R
- Interaction effects in network meta-analysis (multiple treatment comparison) in R?
- Multivariate meta-analysis in R: how to investigate network of variables
- Multivariate meta analysis and metaregression with multi arm randomized controlled studies
Indeed, mvmeta in Stata and R, metafor in R, and WinBUGS can, for instance, be used to perform such analysis. However, it is unclear whether any of the above approaches is better than the others.
Accordingly, my question is mainly comparative. In other words, which is the most potent and accurate of the available approaches for network meta-regression?
