Méta-analyses en réseau : intérêt et limites en oncologie

In the last decade, a new method has emerged called 'network meta-analysis' to take into account all randomized trials in a given clinical setting to provide relative effectiveness between different treatments, whether or not they have been compared (pairwise) in randomized controlled trials. Network meta-analyses combine the results of direct comparisons from randomized trials with indirect comparisons between trials (i.e. when two treatments were not compared with each other, but have been studied in relation to a common comparator). The purpose of this note is to explain this method, its relevance and its limitations. A worked example in non-metastatic head and neck cancer is presented as illustration.