Informing treatment decisions through meta-analysis: to network or not?

Abstract

Evidence from randomised trials and their meta-analyses is typically formed of head-to-head comparisons of a couple of treatments; multiarm trials are infrequent. However, in real-life healthcare, there are many more than two treatment options for a particular condition. To be relevant for the shop-floor of practice, evidence-based medicine requires the use a comprehensive approach to compile, compare and contrast evidence on all options in one synthesis. Network meta-analysis (NMA) offers exactly such a solution. It generates a rank order of the available treatments for practitioners and policymakers that has the merit of objectivity. However, reviewers should make transparent the limitations of NMA as it uses direct and indirect comparisons, inevitably collating data with a certain degree of heterogeneity. This approach is increasingly being deployed to underpin evidence syntheses for incorporating research into practice.