Design and Analysis of Clinical Trials with Time-to-Event Endpoints

摘要

This publication is a collection of articles/contributions on methodologies specifically related to time-to-event problems that involve (right) censored data and arise in health research. The publication comprises some 22 chapters, with the first nine focusing on statistical methods and the latter ones more oriented to problems with specific diseases (oncology, cardiovascular antibiotics etc.). The first chapter introduces time-to-event designs via a short discussion of Phase II and a more detailed examination of Phase III clinical trial designs. The chapter examines issues related to sample size considerations, accrual rates and follow-up duration (including losses to follow-up) as well as methods of comparison such as the logrank test. While comprehensive, some contemporary aspects of time-to-event issues such as the impact of cure (which can occur with some types of cervical cancers) are not discussed. Such cases can have a profound impact on sample size calculations. The treatment of parametric models in Chapter 2 is somewhat superficial, and some discussion and guidelines on the selection of appropriate parametric models would have been useful. Some discussion of normally censored data (as one might obtain by transforming survival times) and censored multiple regression models would also have been of value. Much of the content is well discussed in standard texts, and some synthesis of model choice and interpretation would be refreshing.