Statistical properties and use of sequential methods in randomized clinical trials when the response criterion is censored.