Estimation of treatment effect under non‐proportional hazards and conditionally independent censoring

In clinical trials with time-to-event outcomes, it is common to estimate the marginal hazard ratio from the proportional hazards model, even when the proportional hazards assumption is not valid. This is unavoidable from the perspective that the estimator must be specified a priori if probability statements about treatment effect estimates are desired. Marginal hazard ratio estimates under non-proportional hazards are still useful, as they can be considered to be average treatment effect estimates over the support of the data. However, as many have shown, under non-proportional hazard, the 'usual' unweighted marginal hazard ratio estimate is a function of the censoring distribution, which is not normally considered to be scientifically relevant when describing the treatment effect. In addition, in many practical settings, the censoring distribution is only conditionally independent (e.g., differing across treatment arms), which further complicates the interpretation. In this paper, we investigate an estimator of the hazard ratio that removes the influence of censoring and propose a consistent robust variance estimator. We compare the coverage probability of the estimator to both the usual Cox model estimator and an estimator proposed by Xu and O'Quigley (2000) when censoring is independent of the covariate. The new estimator should be used for inference that does not depend on the censoring distribution. It is particularly relevant to adaptive clinical trials where, by design, censoring distributions differ across treatment arms.

[1]  L. J. Wei,et al.  The Randomized Play-the-Winner Rule in Medical Trials , 1978 .

[2]  Bias correction for score tests arising from misspecified proportional hazards regression models , 2001 .

[3]  D. Cox.  Regression Models and Life-Tables , 1972 .

[4]  J. Robins,et al.  Correcting for Noncompliance and Dependent Censoring in an AIDS Clinical Trial with Inverse Probability of Censoring Weighted (IPCW) Log‐Rank Tests , 2000, Biometrics.

[5]  J M Lachin,et al.  The use of response-adaptive designs in clinical trials. , 1993, Controlled clinical trials.

[6]  Daniel L. Gillen,et al.  Nontransitivity in a class of weighted logrank statistics under nonproportional hazards , 2007 .

[7]  P. Marzec,et al.  Generalized martingale-residual processes for goodness-of-fit inference in Cox's type regression models , 1997 .

[8]  Robert H. Bartlett,et al.  Randomized play the winner clinical trials , 1986 .

[9]  R. Gill,et al.  Cox's regression model for counting processes: a large sample study : (preprint) , 1982 .

[10]  John D. Kalbfleisch,et al.  Misspecified proportional hazard models , 1986 .

[11]  J O'Quigley,et al.  Estimating average regression effect under non-proportional hazards. , 2000, Biostatistics.

[12]  Lanju Zhang,et al.  Response‐adaptive randomization for survival trials: the parametric approach , 2007 .

[13]  H. White.  Maximum Likelihood Estimation of Misspecified Models , 1982 .

[14]  Daniel L Gillen,et al.  The separation of timescales in Bayesian survival modeling of the time-varying effect of a time-dependent exposure. , 2008, Biostatistics.

[15]  J. Kalbfleisch,et al.  The Statistical Analysis of Failure Time Data: Kalbfleisch/The Statistical , 2002 .

[16]  L. J. Wei,et al.  The Robust Inference for the Cox Proportional Hazards Model , 1989 .

[17]  E. Shaw,et al.  RSR13 plus cranial radiation therapy in patients with brain metastases: comparison with the Radiation Therapy Oncology Group Recursive Partitioning Analysis Brain Metastases Database. , 2003, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[18]  Theo Stijnen,et al.  Interim analysis on survival data: its potential bias and how to repair it , 2005, Statistics in medicine.