ADDIS: an automated way to do network meta-analysis

In evidence-based medicine, meta-analysis is an important statistical technique for combining the findings from independent clinical trials which have attempted to answer similar questions about treatment's clinical eectiveness [1]. Normally, such meta-analyses are pair-wise treatment comparisons, which only include the comparisons between two treatments, e.g. treatment A and placebo. When additional treatments are of interest (e.g. treatment B and treatment C), pair-wise treatment comparison starts showing its limitations as it only accesses the evidence from direct comparisons between two treatments and can not guarantee consistency between comparisons. Network meta-analysis is a statistical method for combining both direct and indirect evidence from multiple trials in order to obtain a single consistent quantitative synthesis [2, 3, 4]. It enables to detect the heterogeneity among dierent trials comparing the same treatments and inconsistency between direct and indirect evidence. Compared to pair-wise meta-analysis, network meta-analysis is rather difficult to conduct due to the need for analyzing inconsistency, specifying the model, assessing convergence, etc. The purpose of this report is to introduce an automated way to perform network meta-analysis through ADDIS (Aggregate Data Drug Information System).