Article Text

other Versions

Download PDFPDF
Component network meta-analysis in a nutshell
  1. Sofia Tsokani1,
  2. Georgios Seitidis1,
  3. Dimitris Mavridis1,2
  1. 1Department of Primary Education, University of Ioannina, Ioannina, Greece
  2. 2Research Center of Epidemiology and Statistics (CRESS-U1153), Université Sorbonne Paris Cité, Paris, France
  1. Correspondence to Sofia Tsokani, Department of Primary Education, University of Ioannina, Ioannina, Greece; s.tsokani{at}uoi.gr

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Several organisations, such as the WHO, have endorsed network meta-analysis (NMA) as a powerful tool in clinical decision making. NMA is a statistical method, which simultaneously compares multiple (three or more) interventions within a single framework, by synthesising direct and indirect evidence from multiple studies, addressing the same scientific question.1–4 Healthcare interventions can be complex/multicomponent in the sense that they consist of multiple, possibly interacting, components. While NMA focuses on estimating intervention effects, component NMA (CNMA) disentangles the effect of each component. Subsequently, CNMA uses these estimates to reconstruct multicomponent intervention effects.5–7 We aim to briefly introduce readers to CNMA and highlight its advantages and limitations.

(Standard) NMA

Suppose we have a set of trials comparing multiple interventions (more than two) for their efficacy/safety, forming a network of evidence. An example of such a network is depicted in the network plot in figure 1. Nodes represent interventions and edges represent direct evidence; interventions are linked with an edge, when there are studies comparing them. Intervention effects between any pair of interventions are informed both from studies directly comparing these interventions (direct evidence) and from other studies (indirect evidence), as long as there is a path one can follow from one node to the other. For example, in the network described in figure 1, there is no line/edge (direct evidence) comparing interventions ‘A+B’ and ‘A+B+C’, but one can go from ‘A+B’ to ‘A+B+C’ via usual care (UC) or another path. By synthesising both direct and indirect evidence, NMA results in more precise effect estimates and allows us to estimate the relative efficacy/safety between any pair of interventions, even of those not compared directly in a study.1–4 The main assumption made is that of transitivity, suggesting that the distribution of effect modifiers should be similar across treatment comparisons. Suppose …

View Full Text

Footnotes

  • Twitter @STsokani

  • Contributors ST drafted and edited the article. GS edited the article. DM conceptualised, edited and revised the article.

  • Funding This project has received funding from the European Union’s Horizon 2020 Research and Innovation Programme (grant agreement No 754936).

  • Disclaimer The funder had no role in research design, decision to publish, or preparation of the manuscript.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.