Potential Benefit of Combining Direct and Indirect Evidence:

1
Potential Benefit of Combining Direct and
Indirect Evidence
Mixed treatment comparisons.

• Deborah M. Caldwell A.E. Ades
• MRC HSRC
• University of Bristol, UK
• www.hsrc.ac.uk/

2
Mixed Treatment Comparisons
Smoking Cessation interventions (Hasselblad,
1998).
A No therapy B Self help C Individual
counselling D Group counselling
3
Objectives.

• Conventional approach focuses on reliability of
  pooled efficacy estimate. But at the expense of
  uncertainty ?
• Potential benefit of combining direct and
  indirect evidence - especially precision.
• Potential role and scope of combining indirect
  and direct data in evidence synthesis.
• Decision-making context

4
What's the potential benefit of indirect data in
reducing uncertainty?

• Compare
• Direct, head-to-head A vs B studies.
• Direct plus indirect evidence involving third
  comparators A vs C, B vs C and A vs D, B vs D.
• Direct plus Indirect plus unrelated,
• apparently irrelevant studies C vs D

• Standard error of treatment effect as the measure
  of uncertainty.

5
Random effects Meta-analysis of A vs B A
Regression Type Model
6
Extension to Mixed Treatment Comparisons
7
Song et al (2003) BMJ 326 472
8
Percentage sd reduction.

9
Role of irrelevant AC studies in BD treatment
comparison.
10
(No Transcript)
11
Probability each treatment is best.
12
Early thrombolysis for the treatment of acute
myocardial infarction.

• Streptokinase, Alteplase, Reteplase
  Tenecteplase.
• Without MTC analysis, series of pairwise
  comparisons.

SK is as effective as t-PA Tenecteplase as
effective as acc t-PA Reteplase at least as good
as SK
SK as good as t-PA? Reteplase as good as acc.
t-PA?
Tenecteplase better than SK? Reteplase as good as
tenecteplase?
13
Summary

• Combining direct and indirect evidence in MTC has
  potential to reduce uncertainty in effectiveness
  estimates.
• Even including irrelevant randomised
  comparisons has reduced uncertainty.
• Question how correct is this procedure ?
• Dont know but seems better than assuming
  indirect comparisons are totally irrelevant.

14
Best available v All available

• Need to summarise/ understand vast amount of
  available evidence.
• MTC has potential to reduce uncertainty
• Synthesis of all available evidence should be
  considered more often.
• MTC offers a single consistent summary of all
  treatment comparisons.

15
Potential Benefit of Combining Direct and
Indirect Evidence.

• Deborah M. Caldwell and A.E. Ades
• MRC HSRC
• Department of Social Medicine
• University of Bristol, UK
• www.hsrc.ac.uk/