Network Meta Analysis in R part II

In this part i am discussing the continuation from the Network Meta-analysis part I. I created this one as a Linkedin article will all the codes and steps explained there. So you may find the tutorial using the link bellow. Here i will add some additional perspectives too.

The focus is on estimating the Network effects using the 'netmeta' R package. While many think that actual NMA comparisons are done using the network graphs, they are actually done using the forest plot. In this tutorial i show how to estimate the network effects and use the Network Meta-analysis forest plots for comparative meta analysis of treatments.

This may seem counterintuitive to many, that the Network Meta-analysis is best evaluated using the Forest plot. But it is the true and i will explain why it makes sense. Network Meta-analysis should be observed more like method, a set of concepts and not a graph such as Network Graph. Forest plots can enable us to accuratelly represet multiple comparisons, network effects, indirect and direct effects separated which would be very difficult using the network graph.

For this reason, the network graphs are mainly used to describe the sample in terms of number of studies or subjects and describe the presence or absence of the direct comparisons in studies. Network graphs can then be combined with the forest plots but the forest plot which explain the network itself and show which comparisons were the best.

A reminder : Network effects contain both the direct and indirect effects and are the highest form of evidence in Clinical Biostatistics.

You may read the full tutorial here: https://www.linkedin.com/pulse/network-meta-analysis-r-part-ii-effects-forest-plots-darko-medin-qvbef/

The Network Meta-analysis in R part 3 will also be out soon! The main topic will be separating the indirect and direct evidence from the network.