Most Cochrane Reviews present comparisons between pairs of interventions (an experimental intervention versus a comparator) for a specific condition and in a specific population or setting. However, it is usually the case that several, perhaps even numerous, competing interventions are available for any given condition. People who need to decide between alternative interventions would benefit from a single review that includes all relevant interventions, and presents their comparative effectiveness and potential for harm. Network meta-analysis provides an analysis option for such a review. Any set of studies that links three or more interventions via direct comparisons forms a network of interventions. Network meta-analysis combines direct and indirect estimates across a network of interventions in a single analysis and can inform comparative effectiveness of multiple interventions.
Care needs to be taken using this method because it is more statistically complex than a standard meta-analysis. In addition, as network meta-analyses generally ask broader research questions, they usually involve more studies at each step of systematic review, from screening to analysis, than standard meta-analysis. It is therefore important to anticipate the expertise, time, and resource required before embarking on one. Network meta-analysis is only valid when studies comparing different sets of interventions are similar enough to be combined. When conducted properly, it provides more precise estimates of relative effect than a single direct or indirect estimate. Network meta-analysis also allows the estimation of the ranking and hierarchy of interventions. Much care should be taken when interpreting the results and drawing conclusions from network meta-analysis, especially in the presence of incoherence or other potential biases.
Cochrane reviewers interested to undertake a network meta-analysis should prepare their protocol according to the guidance in Chaimani et al. and their final report according to the PRISMA extension statement by Hutton et al.