|
Abstract:
|
Survival is the gold standard of efficacy measures for oncology clinical trials, however, the durability of response is increasingly used as an endpoint of clinical benefit, especially in some chronic indications such as chronic myeloid leukemia. The ITT duration of response, defined by including the non-responders as having the worst possible duration of response, i.e., zero, contains information about the response rate and the durability of response among responders. It can be used to make comparisons between two treatment arms. In this article, we compare the performance of the log-rank statistic to those of parametric and nonparametric tests based on the expected duration of response and explore the methods using Monte Carlo simulations. Our results show that the log-rank test, though not optimal in all situations has some desirable properties.
|
