This workshop is given on behalf of the Statistical Methods Group.

Workshop type: Training.

Objective: To describe and discuss methods for multiple treatment meta-analyses, which allow determination of the 'best' among a selection of interventions.

Summary:
Systematic reviews focus on pair-wise, direct comparisons of treatments which can make it difficult to determine what is the ‘best’ treatment when there are multiple treatment options. In the absence of a collection of large, high-quality randomized controlled trials (RCTs) comparing all treatments, the comparison of multiple treatments involves the use of indirect comparisons. For example, without RCTs directly comparing treatments A and B, an indirect estimate of the benefit of A over B may be obtained by comparing trials of A vs C with trials of B vs C. In the general case, we have to deal with a set of treatments that we aim to rank a ccording to their effectiveness, and every study included in a meta-analysis may contribute partly to the evidence. This setting is referred to as multiple treatment meta-analysis (MTM). The workshop will provide an introduction to the concepts and methodology of MTM. We will outline the benefits of combining direct and indirect information, such as greater precision and ranking of the treatments compared to conventional, head-to-head meta-analysis. We will explore the different implementation alternatives (Bayesian versus frequentist) through two worked examples of multiple treatment meta-analyses. We will discuss some specific methodological issues, such as the underlying assumptions of the MTM models and evidence consistency. We will also address concerns regarding bias due to confounding. The workshop will close with an open debate on the policy of The Cochrane Collaboration regarding MTM analysis.

Intended audience: Review authors.

Technical level expect ed of participants: Intermediate/Advanced.