Home Page >
Research areas >
Methodology >
Applied stat methodology >
Developing more flexible and informative methods to analyse survival data
Developing more flexible and informative methods
to analyse survival data
Time-to-event outcomes are primary to most of CTU’s trials and
other clinical studies. Good estimates of the hazard function
assist scientists in interpreting study results critically, and
give insight into disease mechanisms, helping to guide future
strategies to be evaluated in CTU trials.
Also, it is often essential to extrapolate beyond the maximum
follow up time observed in the trial. This is particularly
important for health economic analyses, where estimates of
long-term benefits and costs of treatments are needed.
Standard methodology (e.g. the Cox model) does not provide such
estimates. Neither does it help us with treatment effects showing
non-proportional hazards. For example, targeted agents in cancer
trials may show non-proportional hazards of the treatment effect on
outcome, because the effect “wears off” after the end of
treatment.
Better theoretical and practical approaches to the analysis of
such data are needed. Royston and Parmar have proposed and
illustrated a whole new range of flexible survival models and we
are currently extending the methods to relative survival and to
patient cure. We are also developing methods to handle
non-proportional hazards in trials and observational studies, with
associated software
developed and made publicly available in Stata.
Key projects
- More complete and clinically
informative methods for analysing non-proportional hazards
data: further research and dissemination of a range of
models that give better descriptions of survival data, and draw
more reliable, robust and informative conclusions than existing
standard methods allow
- Applications of flexible parametric
modelling to survival data in trials and observational
studies, including the possibility of using the models to
extrapolate survival times in health economic evaluations
Selected
publications
- Royston P, Parmar MKB, The use of restricted mean survival time
to estimate the treatment effect in randomized clinical trials when
the proportional hazards assumption is in doubt. Statistics in
Medicine 2011, DOI: 10.1002/sim.4274
- Lambert P, Royston P. Further development of flexible
parametric models for survival analysis. Stata Journal 2009;
9:265-290
- Sauerbrei W, Royston P, Look M. A new proposal for
multivariable modelling of time-varying effects in survival data
based on fractional polynomial time-transformation. Biometrical
Journal 2007; 49:453-473
