We propose novel methods for the analysis of survival data arising from studies of organ failure patients. The organ failure setting yields data structures to which existing univariate and multivariate survival analysis methods cannot be applied to address many questions of great interest to clinicians, health policy makers and patients. The issues addressed by the proposed methods are general, in that they will have wide applicability outside the organ replacement setting. There is currently great interest in quantifying the survival benefit of organ transplantation and, correspondingly, this proposal is targeted strongly in that direction.
In Aim 1 we propose semiparametric methods to quantify the effect of an experimental time-dependent therapy relative to conventional therapy on mean time until failure. Methods developed in Aim 1 will be used to quantify the benefit of expanded criteria donor kidney transplantation relative to conventional therapy.
In Aim 2, we develop semiparametric methods to predict patient-specific differences between post-treatment and treatment-free mean lifetime. These predictions are intended to be the basis of a score to replace the current system for prioritizing patients wait listed for liver transplantation. We also propose methods for estimating the marginal effect of a time-dependent treatment on mean lifetime, obtained by averaging over the patient- specific estimates used in the proposed score.
In Aim 3, we develop and evaluate semiparametric methods to make covariate-adjusted comparisons of gap time survival distributions. The methods proposed in Aim3 will be used to compare mean lifetime of first and second kidney transplants, in order to assess the value of repeat kidney transplantation.
We propose novel methods for the analysis of survival data arising from studies of organ failure patients. There is currently great interest in quantifying the survival benefit of organ transplantation and, correspondingly, this proposal is targeted strongly in that direction.
Showing the most recent 10 out of 66 publications
