Key points are not available for this paper at this time.
Carrent methods for adjusting estimated survival curves for the effects of important prognostic factors are reviewed. These methods often are not useful because of the censoring pattern or because the underlying assumptions are not met. A new method, which is a generalization of the product-limit estimator, is proposed and its properties are considered. It also has limitations, but is useful in several cases where the existing methods are not suitable.
David A. Amato (Fri,) studied this question.