Network meta-analysis (NMA)
Information on best practice for conventional indirect comparisons and network meta-analysis (NMA) is summarised on this page, with links to useful resources.
For more information describing NMA see Overview of evidence synthesis and NMA . The GetReal review on NMA methods can be found at Efthimiou et al., 2016 and the articles identified in this review can be found at this Zotero page for Network Meta-analysis.
Assessing the assumptions of NMA
NMA adopts the same set of assumptions as a usual (pairwise) meta-analysis, but also uses an additional assumption, called transitivity (also called similarity or exchangeability) which may be hard to assess (Ades, 2011; Salanti, 2012; Efthimiou et al., 2016).
- Transitivity assumes that information for the comparison between treatments B and C can indeed be obtained via another treatment, A, using the direct comparisons for A vs. B and A vs. C. Researchers can assess this assumption by checking the distribution of effect modifiers across comparisons. See ISPOR Guidance on this: Jansen et al., 2011.
- Other considerations include checking whether the missing treatments in each trial are ‘missing at random’ or checking that the choice of treatment comparisons in the trials is not associated either directly or indirectly with the relative effectiveness of the interventions and that the treatments in the network are ‘jointly randomisable’ (Salanti, 2012; Lu & Ades, 2006; Cipriani et al 2013).
- When the transitivity assumption does not hold, this may lead to inconsistencies in the data. Inconsistency refers to the statistical disagreement between the observed direct comparisons and the comparisons based on indirect sources of evidence. Checking the network for inconsistencies offers a method for assessing the validity of the transitivity assumption.
- For ISPOR’s position on NMA, see Ades, 2011.
Implementing an NMA
There are a range of approaches that can be used to conduct an NMA, using either frequentist or Bayesian methods. More details, including information on models and software codes, can be found in:
- International Society for Pharmacoeconomics and Outcomes Research (ISPOR) reports on interpreting and conducting indirect treatment comparisons (Jansen et al., 2011; Hoaglin et al., 2011)
- NICE Decision Support Unit documents on evidence synthesis for decision-making (NICE DSU TSD 1-6).
- Different approaches for assessing the consistency of a network include:
- the loop-specific approach
- the node-splitting approach
- the Lu & Ades inconsistency model
- the design-by-treatment inconsistency model
- the Q-statistic for inconsistency in NMA.
- The GetReal review on NMA methodology (Efthimiou et al., 2016) gives a general review of methods for implementing an NMA model and checking for inconsistencies in the network. It also covers more advanced issues in NMA, such as:
- the use of different effect measures
- extensions of NMA to account for effect modifiers
- investigating potential bias in NMA
- modelling multiple outcomes
- definition of nodes in NMA
- planning future studies.
Other resources for conducting and appraising an NMA
- Cochrane Methods group for ‘Comparing Multiple Interventions’ provides protocol template for NMA (see Cochrane Method group).
- An ISPOR checklist of good research practices for conducting and reporting a NMA (Hoaglin et al., 2011).
- The PRISMA extension statement offers guidelines, aiming to improve the reporting of NMA (Hutton et al., 2015)
- A simplified ISPOR checklist to assist decision makers in evaluating a NMA (Jansen et al., 2011)
- Detailed approaches to evaluate the quality of evidence from a NMA are described in Puhan et al., 2014 and Salanti et al., 2014.
Orestis Efthimiou, University of Bern and University of Ioannina School of Medicine