Indirect and mixed treatment comparisons in arthritis research.

Abstract

Evidence for the efficacy of biologic therapies in inflammatory arthritis comes overwhelmingly from placebo-controlled trials. Increasingly, however, authorities responsible for purchasing and reimbursement have tried to determine whether there are differences between these powerful new therapies, which would lead them to recommend some in preference to others, either on grounds of efficacy or cost-effectiveness. In the absence of head-to-head trial comparisons, indirect comparisons may be used. Furthermore, network meta-analysis, also known as mixed treatment comparisons can combine information from trials in a connected network. These methods allow inferences about head-to-head comparisons even when there is little or no head-to-head evidence, which has caused some concern. In this article we briefly review these methodologies and describe results from recent applications to inflammatory arthritis in the clinical literature. We then focus on how the methodologies are used in decision making, taking as an illustration some recent technology appraisals conducted by the National Institute for Health and Clinical Excellence in the UK. We conclude that, in practice, the key decisions have been based on results from placebo-controlled trials.