Emergent approaches to the meta-analysis of multiple heterogeneous complex interventions.

G J Melendez-Torres,James Thomas,Chris Bonell

doi:10.1186/s12874-015-0040-z

Abstract

BackgroundMultiple interventions meta-analysis has been recommended in the methodological literature as a tool for evidence synthesis when a heterogeneous set of interventions is included in the same review—and, more recently, when a heterogeneous set of complex interventions is included. However, there is little guidance on the use of this method with complex interventions. This article suggests two approaches to model complexity and heterogeneity through this method.Discussion‘Clinically meaningful units’ groups interventions by modality or similar theory of change, whereas ‘components and dismantling’ separates out interventions into combinations of components and either groups interventions by the combination of components they demonstrate or extracts effects for each identified component and, possibly, interactions between components. Future work in systematic review methodology should aim to understand how to develop taxonomies of components or theories of change that are internally relevant to the studies in these multiple interventions meta-analyses.SummaryDespite little meaningful prior guidance to its use in this context, multiple interventions meta-analysis has the potential to be a useful tool for synthesising heterogeneous sets of complex interventions. Researchers should choose an approach in accordance with their specific aims in their systematic review.

Highlights

Multiple interventions meta-analysis has been recommended in the methodological literature as a tool for evidence synthesis when a heterogeneous set of interventions is included in the same review—and, more recently, when a heterogeneous set of complex interventions is included
Summary: Despite little meaningful prior guidance to its use in this context, multiple interventions meta-analysis has the potential to be a useful tool for synthesising heterogeneous sets of complex interventions
Researchers should choose an approach in accordance with their specific aims in their systematic review

Summary

Introduction

Multiple interventions meta-analysis has been recommended in the methodological literature as a tool for evidence synthesis when a heterogeneous set of interventions is included in the same review—and, more recently, when a heterogeneous set of complex interventions is included. Known as network meta-analysis, mixed treatment comparison and multiple treatments meta-analysis, the benefits of using multiple interventions meta-analysis (the term we use throughout this discussion) are several: it allows for the inclusion of all relevant evidence, including trials testing one active intervention against another and each relevant arm in trials with multiple arms; it allows head-to-head comparisons of distinct interventions; and when implemented with a Markov chain Monte Carlo method, it allows interventions to be ranked probabilistically by their relative efficacy [9, 10] These methods are gaining increasing popularity and Melendez-Torres et al BMC Medical Research Methodology (2015) 15:47 have been suggested for use with complex interventions [11]. In all of this work, multiple interventions meta-analysis was primarily considered for use with pharmacological, ‘non-complex’ interventions

Discussion

Conclusion