Network meta-analysis: Integrating results of direct and indirect treatment comparisons

Chairs: N Schooler, G Haynes, M Stewart

Meta-analysis is a statistical technique widely used to integrate results of studies comparing the same treatments in order to obtain the best estimate of effect size. Network meta-analysis extends this methodology to comparisons of treatments that are not included in the same studies, i.e., have not been directly compared to each other. This potentially allows a wider range of comparisons to be considered in estimates of effect size; and may also be particularly useful as the results of trials with new interventions must be integrated with results of earlier comparisons. This session will present an overview of the statistical concepts used in network meta-analysis, and examples based on antipsychotics in schizophrenia and mood stabilizers in bipolar disorder will be used.  Commentaries by academic and industry clinical trialists, a statistician, and a representative from the payer community, will address the strengths and limitations of this new approach and how it is viewed by the field.