6Adverse Events and Other Sparse Outcome Data

6.1 Introduction

This chapter focuses on the specific challenges for pairwise and network meta-analysis of sparse outcomes (i.e., low numbers of events) and few data (i.e., small numbers of studies; especially a lack of multiple estimates for specific comparisons). Sparse outcomes are predominantly, but not exclusively, encountered with adverse event outcomes. In addition to evaluating efficacy, network meta-analysis is increasingly being used to synthesise information on adverse events relating to the interventions of interest. The aim of such an analysis could be to explore concerns regarding the safety of a drug or to quantify the risks and problems associated with an intervention. This could potentially inform either a clinical decision model weighing up the benefits and risks associated with an intervention strategy (Sutton et al., 2005; Braithwaite et al., 2008) or an economic decision model of the form discussed in Chapter 5. We consider the synthesis of RCT adverse event data and the unique technical challenges a quantitative synthesis of such data presents, although we acknowledge that RCT data alone may not always be the optimal choice of data to use (Loke et al., 2011). The predominant issues considered relate to the sparseness of such data and specifically to where one or more arms of trials observe zero events (and similarly for very common events where everyone in a trial arm has an event). For a more general coverage ...

Get Network Meta-Analysis for Decision-Making now with the O’Reilly learning platform.

O’Reilly members experience live online training, plus books, videos, and digital content from nearly 200 publishers.