Bayesian Evidence Synthesis

Evidence synthesis involves the development of techniques to combine multiple sources of quantitative evidence. In health technology assessment, meta-analysis is a well-established body of techniques for combining evidence from high-quality trials.

Recently, a number of researchers have been developing methods, grounded in Bayesian statistical theory, to complement and enhance conventional meta-analysis. These methods provide ways of tackling many of the challenges that arise in evidence synthesis, such as heterogeneity, indirect comparisons and baseline risk effects. By explicitly including study quality in the synthesis, these methods are able to draw on a broad evidence base to support decision-making.

Bayesian evidence synthesis is an area of growing interest for HEDS through its link with CHEBS (the Centre for Health Economics and Bayesian Statistics). Current work in this area includes:

• An investigation of modelling methods for integrating routine and trial data in the evaluation of cancer screening programmes (this work is being carried out by Jason Madan for his PhD, funded by a Researcher Development Award from the Department of Health).
• A meta-regression of treatments for Rheumatoid Arthritis (with Dr. Richard Nixon of the MRC Biostatistics Unit).

HEDS also has strong research interests in Expected Value of Information analysis (see CHEBS research page), and Expert Elicitation (BEEP), both of which are related to Bayesian evidence synthesis.