Session: Survival Analysis
Time-to-event outcome measures are used in
many trials. Some examples of such measures are time to disease
recurrence, time to relief of symptoms and time to death. This
session will consider newer methods for the design and analysis of
such trials, with particular emphasis on the analysis. There are
now some ‘standard’ tools to analyse such data which include the
logrank test, hazard ratio, Kaplan-Meier curves and Cox Models.
However, a number of these tools are used uncritically and make
assumptions, such as proportional hazards. We also need
alternatives or extensions of these methods for dealing with more
complex situations, for example when there is unplanned crossover
in the trial. A number of extensions of, and alternatives to,
these methods have been proposed when assumptions cannot be assumed
to hold, and when things are more complex.
Aims
- present some of these methods looking at
them critically both from a methodological and practical
perspective, with the aim of proposing a broader range of practical
methods for routine use in trials.
Oral Presentations
Click on the links below to view the presentations.
Survival
analysis: coping with non-proportional hazards in randomized
trials
Evaluation of
methods that adjust for treatment switching in clinical
trials