"...Network meta-analysis (NMA) is a method used to assess the comparative effectiveness of experimental treatment(s) among similar patient populations that have not been compared directly in randomized trials."
Network meta-analysis (NMA) (a novel approach to pooling data from randomized controlled trials) has grown in popularity and may be a solution for clinicians seeking to compare more than two treatments for specific diagnoses or diseases. In addition to comparing different treatments simultaneously, network meta-analysis combines data from direct evidence (relevant head-to-head trials) and indirect evidence (Hutton, 2013). Further NMA is a statistical term that refers to multiple treatment comparisons, and direct or indirect comparisons for a single analysis. This method results in a summary of estimates of efficacy /or safety of treatments that may not have been otherwise possible (known as a mixed treatment comparison methodology). Li et al (2011) said "...network meta-analysis, in the context of a systematic review, is a meta-analysis in which multiple treatments (three or more) are compared using direct comparisons of interventions within randomized controlled trials and indirect comparisons across trials based on a common comparator."In September 2014, Li et al said that the "...network meta-analyses can be improved by searching more sources ...and by involving a librarian".
Statistically, the NMA extends the traditional meta-analyses (where all included studies compare the same intervention and same comparator) by including multiple different pair-wise comparisons across a range of interventions. With the NMA, the relative efficacy (or safety) of an intervention versus competing interventions is obtained in the absence of head-to-head evidence; an indirect comparison of two interventions is made via a common comparator. Further, the NMA strengthens (arguably) the inference by including direct and indirect effects. Bayesian approaches to the NMA are considered the gold standard as they allow for probabilistic interpretations. The ranking of interventions regarding their ability to provide the best outcome is particularly useful for clinicians.
Challenges & benefits
In the absence of clinical trials involving a direct comparison of treatments of interest, an indirect comparison can provide useful evidence of the difference in treatment effects among competing interventions (which otherwise would be lacking) and for judiciously selecting the best choice(s) of treatment.
According to Caldwell et al (2010), NMA can be used in conjuction with summaries from systematic reviews; moreover, mixed methods can be used in overviews to provide a single coherent analysis of all treatment comparisons and to check for evidence consistency.
The challenge in using the NMA method vis a vis traditional meta-analysis is that it is more likely to be valid when analyzing very similar studies for very similar patient populations.
NMA extends the number and types of studies being combined, so there's potential for combining studies that are not similar; the quality of some recent network analyses in the hypertensive literature highlights the problems of this type of analysis.
According to White et al (2011), "...network meta-analysis (multiple treatments meta-analysis, mixed treatment comparisons) attempts to make the best use of a set of studies comparing more than two treatments. However, it is important to assess whether a body of evidence is consistent or inconsistent."
Mixed treatment comparisons, of which NMA is a special type, use direct and indirect evidence for particular pairwise comparisons, thereby synthesizing a greater share of the available evidence than a traditional meta-analysis.
It may be argued that indirect comparisons are essential when direct comparisons are unavailable; however, it is important to realize both direct and indirect evidence contributes to the total body of evidence. Results from indirect evidence combined with direct evidence strengthens the assessment between treatments
In 2013, Bafeta et al found that electronic search strategies in 87 (72%) network meta-analyses no literature search was reported; the report search was only in one database (did not search other sources) or did not report an assessment of risk of bias.
ITC software – application for performing indirect comparisons
ITC software is an application suitable for Windows that uses the ‘Bucher method’ to perform indirect comparisons. It requires as input the mean relative treatment effect and the respective 95% confidence interval for each direct comparison involved in a specific indirect root. The application can incorporate estimates on different effect measures (odds ratio, risk ratio, risk difference, mean difference, hazard ratio).
The mvmeta command in STATA employs a recent approach to network meta-analysis that handles the different treatment comparisons appeared in studies as different outcomes. The command can perform fixed and random effects network meta-analysis assuming either a common or different between-study variances across comparisons. Both consistency and inconsistency models (the ‘design-by-treatment model’ or ‘Lu & Ades model’) have been implemented as well as network meta-regression models that can incorporate covariates. The command contains also an option that enables the estimation of ranking probabilities.