Identification of the Minimum Effective Dose for Right-Censored Survival Data

Abstract

The problem of identifying the minimum effective dose (MED) is considered when the survival data observed in a dose-response study are subject to random right-censorship, where the MED is defined to be the lowest dose level producing a better survival over that of the zero-dose control. Proposed herein is a class of testing procedure based on different types of weighted logrank statistics incorporated into the step-down closed testing scheme suggested by Tamhane et al. (1996). The associated p-value of the identified MED is also obtained. The comparative results of a Monte Carlo error rate and power study for small sample sizes are presented and discussed. An appropriate numerical example is finally reported.