Home Page >
Research areas >
Methodology >
Applied stat methodology >
Modelling to identify prognostic and predictive patient factors
Modelling to identify prognostic and predictive patient
factors
In clinical trials, knowledge of prognostic
factors helps trialists to recruit appropriate patients and allows
stratification by disease severity. In meta-analysis,
prognostic information allows results to be validly combined across
trials.
Reliably identifying factors which predict
response to, or toxicity from, a particular therapy is central to
the aim of modern “personalised” or “stratified” medicine, allowing
treatment to be targeted at those patients most likely to
benefit.
The CTU has been developing methods to
determine sample size for multivariable prognostic studies
(initially based on the D statistic developed by Royston and
Sauerbrei) and is developing methods for external validation of Cox
models and flexible parametric survival models. We are also
investigating simple ways such as truncation, to make modelling of
continuous prognostic and predictive factors more reliable, as well
as further developing modelling through fractional polynomials. We
have developed freely available statistical software in Stata.
The Unit has access to the valuable resource
of past and present CTU trials, which will provide a unique
test-bed for new methodology.
Key projects
- Design of multivariable prognostic
studies: design issues, including sample size
- External validation: define
good validation; develop, test and implement practical methodology
for validation of prognostic/predictive survival models in
independent data
- Multivariable modelling methods,
improved models: further develop techniques for reliable
and accurate prognostic and predictive assessments in cancer and
other diseases. Development of more reliable models. Includes
reanalysis of older CTU trials from a prognostic and predictive
perspective
Selected publications
- Choodari-Oskooei B, Royston P, Parmar MKB. A Simulation Study
of Predictive Ability Measures in a Survival Model I: Explained
Variation Measures. Statistics in Medicine 2011; 30(34) (DOI:
10.1002/sim4242)
- Altman DG, Vergouwe Y, Royston P, Moons KG. Prognosis and
prognostic research: validating a prognostic model. BMJ 2009;
338:b605
- Royston P, Sauerbrei W. Two techniques for investigating
interactions between treatment and continuous covariates in
clinical trials. Stata Journal 2009; 9:230-251
- Royston P, Moons KG, Altman DG, Vergouwe Y. Prognosis and
prognostic research: Developing a prognostic model. BMJ 2009;
338:b604
- Sauerbrei W, Royston P, Binder H. Selection of important
variables and determination of functional form for continuous
predictors in multivariable model-building. Statistics in Medicine
2007; 26:5512-5528
- Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous
predictors in multiple regression: a bad idea. Statistics in
Medicine 2006; 25:127-141
