This chapter provides background information about indirect treatment comparisons and an introduction to the methodology behind the ITC application.
1.1.2 In this chapter
1.2 Indirect Treatment Comparison
When comparing treatment interventions, the need for indirect approaches is increasing. A direct assessment of interventions A and B is available if a randomized controlled trial of A versus B has been conducted. However, many competing interventions have not been compared directly and/or such direct evidence is limited and insufficient. The reasons for lack of a direct comparison vary.
More complex indirect evidence settings can arise. In the next simplest setting, we may have direct evidence from A versus C, B versus D, and C versus D. Using this evidence, we can attempt an indirect comparison of A versus B using, in particular, the direct evidence of C versus D. Even in the situation of A versus C, B versus C, D versus C, and D versus F, treatment F can be an important contributor to the indirect comparison of A versus B. The web of direct and indirect evidence can be complex.
Within this web of evidence, there is often a need to synthesize evidence from randomized controlled trials, and methods for deriving indirect treatment comparisons using meta-analysis are of prime interest.
The Bucher et al. (1997) method has been widely used for making indirect comparisons. The approach is pragmatic, and the assumption of independence among trials often holds in settings where the direct comparison between treatments is not available and one needs to use results (possibly from meta-analysis) from non-overlapping treatment pairs. These indirect approaches have recently been applied in published meta-analyses by Yazdanpanah et al. (2004) and Lim et al. (2003).
For discrete outcomes, expanding the indirect odds ratio approach by Bucher et al. (1997) for more complex webs of evidence involving any number of direct comparisons was considered. This generalized approach was then considered for the relative risk (RR), hazard ratio, risk difference, and mean difference. The ITC software application has been developed to assist analysts in applying this expanded approach.