Network meta-analysis

Abstract

Clinicians are often faced with results of randomised controlled trials comparing different interventions for a condition. However, selecting from a range of interventions is difficult when no head-to-head trials are comparing their safety and efficacy. Network meta-analysis (NMA) extends the principles of conventional meta-analysis to allow assessment of multiple treatments in a single analysis. Hence it is also called as multiple treatment meta-analysis or mixed treatment comparisons. Importantly NMA can provide evidence on 'relative ranking' of multiple interventions. The ability to synthesise indirect evidence and evaluate multiple interventions with a common comparator in one analysis separates NMA from conventional pairwise meta-analyses. NMA is vital for evidence-based decision-making because it allows assessment of direct as well as indirect evidence. Given its complexity compared with conventional meta-analyses, the involvement of both subject experts and experienced biostatistician is necessary when planning an NMA. This is particularly important because crucial judgements and assumptions are involved. This chapter briefly covers the principles of NMA.