Grup de Recerca en Anàlisi Estadística de la Supervivència. GRASS

Xerrada de Urania Dafni (Ranny)

Seminari GRASS - Dijous 4 de novembre de 2010

 

Lloc: Edifici C5, PLANTA 0, AULA 16. Campus Nord.

EXPOSICIÓ RANNY
Hora: 11:00
Títol: Estimation of Survival Curves in the Case of “Time Varying Treatment”. Urania Dafni and Dimitris Karlis.
Resum: 
Estimating survival curves in a non-parametric way via the use of Kaplan-Meyer probabilities is common practice. In the clinical trial setting, in the case of randomized treatments, this approach is widely applied and its properties are well understood. In cases where the treatment varies across time (e.g. switching to another therapy) attempts have been made based on the classical Kaplan-Meyer to estimate the survival curve for the group of patients that changed treatment.
For these cases the time varying Cox model can estimate efficiently and without bias the treatment effect, but no information about the survival curves is given. In this talk we aim at filling the gap by considering an appropriate estimate for the survival curve in this case.
We developed an alternative estimate for the survival curve based on similar conditional arguments as in the typical Kaplan-Meyer but taking into account the survival and treatment history of the patients. Properties of the proposed estimator are discussed. Simulation evidence is also provided. The proposed survival estimates allow the extraction of information on the treatment effect for those that changed treatment and hence provide an improved estimate for the differential treatment benefit. We exploit the usage of these curves for deriving more accurate information on the effects of two initially randomized treatments, making use of the whole history of the patients.
We will also discuss the relationship of the new estimator with the Cox time varying model approach and exploit the kind of information we may derive from it. We think that the proposed estimator can supplement the time varying Cox model as it allows its proper visualization.
Real data examples will be discussed as well as the potential of the new approach for improving the design of trials when switching treatment is possible. In this case the existing design ignoring this aspect leads to reduced power at the end of the study.