Home Page >
Research areas >
Methodology >
Applied stat methodology >
Answering questions that cannot be answered by randomised trials
Answering questions that cannot be answered by randomised
trials
Using
randomised clinical trials (RCTs) to address all key
questions about patient management is not feasible. However, with
careful methodology, it may be possible to address more than just
the randomised comparison using the other high quality data
collected in a trial. For example, the impact of second-line
treatments or concomitant medications both of which are highly
relevant to cancer and HIV.
The Unit has completed a major causal
analysis demonstrating substantial reductions in early mortality on
anti-retroviral therapy in DART
trial participants who were taking cotrimoxazole prophylaxis.
CTU trials provide a particularly rich resource to assess the
impact of such aspects of patient management.
Another major role of causal models for CTU
trials is to assess the effect of different parts of the strategies
evaluated, for example how much of the benefits of a
strategy are due to a later switch to more toxic second-line
therapy. Such causal effects are important to inform patient care
and also to identify future strategies for evaluation. Several
observational studies are led from the CTU, and causal methods are
also needed to gain the most from these.
The methodology itself is complex and still
in its infancy, with unanswered questions around the practical
application such as the assessment of model fit and robustness to
assumptions. The CTU will continue to collaborate with researchers
at Harvard, University of California and San Francisco.
Key
projects
- How to compare ‘dynamic’ treatment
regimes: causal modelling to investigate when to start HIV
antiretroviral therapy, based on cohort studies
- Effect of "second-line"
treatments: causal analysis for trials with differential
use of alternative second-line treatments, based on trial data in
HIV and cancer (FOCUS)
Selected
publications
- Bond SJ, White IR, Walker AS.
Instrumental variables and interactions in the causal analysis of a
complex clinical trial. Statistics in Medicine 2007;
26:1473-1496.
- Walker AS, Ford D, Gilks CF, Munderi P,
Ssali F, Reid A, Katabira E, Grosskurth H, Mugyenyi P, Hakim J,
Darbyshire JH, Gibb DM, Babiker AG. Daily co-trimoxazole
prophylaxis in severely immunosuppressed HIV-infected adults in
Africa started on combination antiretroviral therapy: an
observational analysis of the DART cohort. Lancet
2010;375:1278-86.
