Methodological considerations in network meta-analysis

Abstract

Network meta-analysis (NMA), multiple treatments meta-analysis (MTM), or mixed treatment comparison (MTC) provides a flexible and powerful generalization of pairwise meta-analysis traditionally used with SRMA [9] for comparison of multiple interventions simultaneously. As has been explained in numerous publications, [10,11] NMA allows for indirect comparisons of treatments that have not been studied in a head-to-head manner. For example, treatment effects from clinical trials comparing X with respect to Y (in XY trials) and trials comparing Z with respect to Y (in YZ trials) can be then pooled together in NMA to obtain an indirect estimate for indirect comparison between X and Z [12] by means of the common comparator Y. Even if direct evidences for X and Z (via XZ trials) exist, combining them with indirect estimates will strengthen the evidence base. [13] In the mid-1990s, the technique of "adjusted indirect comparisons" was used to compare multiple interventions. This technique was subjected to extension and periodical revisions, leading to the development of NMA. Higgins and Whitehead, [10] Bucher et al. (1998), Lumley (2002), Lu and Ades (2004) have made enormous contributions toward the development of NMA. Network plot is the pictorial representation of all comparisons to be made within a NMA. A well-designed network plot will foster the analysis and will be helpful in procuring reliable estimates. A network plot can either be Network meta-analysis (NMA) is an extension of pairwise meta-analysis that facilitates comparisons of multiple interventions over a single analysis. It is the method in which multiple interventions (that is, three or more) are compared using both direct comparisons of interventions within randomized controlled trials and indirect comparisons across trials based on a common comparator. NMA is methodologically complex compared to simple pairwise meta-analysis as it accounts for a broader evidence base. Results from NMA are more useful to policy makers, service commissioners, and providers when making choices between multiple alternatives than those from multiple, separate pairwise meta-analyses. It can be an ideal choice to be extended to compare complex interventions that are multifaceted. Apart from the numerous benefits the NMA offers, it is prone to methodological complications that need to be understood, implemented, and finally reported correctly. This article is meant to provide a primer to the various methodological issues pertaining to NMA. The NMA can be as valid as a standard pairwise meta-analysis if these methodological issues are taken care of.

Network meta-analysis: a technique to gather evidence from direct and indirect comparisons

“…They have also facilitated simultaneous inference regarding all treatments by potentially ranking these treatments. Calculations of the probability of one treatment being the best or worst for a specific outcome through rank orders or rankograms (graphical methods) are usually employed, and facilitate the interpretation of the results 26,29…”

“…For instance, in closed loop networks both direct and indirect evidence are available and it is assumed that for each pairwise comparison (A-B; B-C and A-C), direct and indirect estimates should be consistent. Violation of these assumptions transgresses the theory of transitivity, where one cannot conclude that C is better than A from trial results that have already proven that C is better than B and B is better than A 26,35 …”

Mapping the characteristics of network meta-analyses on drug therapy: A systematic review

“…In the future, a standardised way of reporting NMA plots and geometry should be considered as an additional parameter of publication reproducibility. The use of rankograms to display the probability to be the best choice among evaluated treatments is a helpful tool for policy makers [ 11 , 18 , 28 ]. However, rankograms alone may not be enough, and graphs would depend on both the nature of data used in the NMAs and the statistical method employed [ 11 , 63 ].…”

“…The indirect comparison method proposed by Bucher et al [ 9 ] provided a potential solution for treatments that have not been directly compared before. However, this model can only be applied to data generated from trials with two arms and with a common comparator, allowing the indirect comparison of three treatments (A vs. B; B vs. C) [ 10 , 11 ]. Thereafter, Lumley [ 12 ] and Lu and Ades [ 13 ] improved indirect treatment comparison techniques, involving more than one common comparator (the linking treatment) and creating NMA, also called mixed or multiple treatment comparison meta-analysis.…”

“…NMA allows to simultaneously combine both direct and indirect results from all studies’ arms into a single pooled effect, which strengthens results and provides a broader picture of all treatments in the same model [ 14 – 17 ]. Moreover, NMAs calculate the probability for each treatment to be the best (or worst) for a specific outcome by creating probability rank orders or rankograms (graphical methods), which are useful for the decision-making process [ 11 , 18 ].…”