Sample size formula for time-to-event data with incomplete event adjudication using generalized logrank statistic.

摘要

Event classification committees (ECC) are used routinely to adjudicate suspected end-points in cardiology clinical trials. When interim analysis are performed in such trials, the final classification for many reported events will not be known, and hence using ECC may introduce delay in adjudication. Cook and Kosorok (2004) proposed a generalized logrank statistic for analyzing time-to-event data with incomplete adjudication. Their results suggest that it may be unnecessary to adjudicate all of the events in a trial. In this research, under the assumption that only a fraction, q, of randomly selected subjects whose events will be adjudicated, we derive the theoretical formula for the variance of the weighted logrank statistic under a general class of local alternatives in which the conditional distribution of time to failure are assumed to vary in the two treatment arms. Using this result, a sample size formula that takes into account adjudication process is derived. Numerical illustrations are provided to evaluate the variance formula via simulating time-to-event data with incomplete adjudication. Numerical studies are conducted to evaluate the power of the logrank test for the time-to-event data with the number of subjects determined on the basis of sample size formula.