|
Abstract:
|
Clinical trials sometimes contain two or more binary response endpoints, such as hematological response, cytogenetic response, and molecular response. Usually, all response rates should be estimated and reported. In a clinical trial setting, these responses are often closely correlated and some patients are not evaluable for one or more responses due to uncontrollable factors, such as insufficient cell quantity in samples leading to indetermination of one or more responses. We propose a self-consistent estimator under the assumption that the response rates are the same for evaluable patients and for inevaluable patients. The proposed estimator is more efficient than the standard estimator in terms of asymptotic relative efficiency. Simulation results and a real example will be provided.
|
