S10 Planning clinical trials to evaluate early detection programs
used IDM. Twenty-nine questionnaires were returned (97%). Responsewas positive with 86% giving a score of “good” to “excellent” for “Overall Satisfact...
used IDM. Twenty-nine questionnaires were returned (97%). Responsewas positive with 86% giving a score of “good” to “excellent” for “Overall Satisfaction”. Based on our user and management experience, IDM provides an attractive option for large multi-center studies. SlO PLANNING CLINICAL TRIALS TO EVALUATE EARLY DETECTION PROGRAMS Ping Hu and Marvin Zelen
Harvard School of Public He&h Boston, Massachuselts Thii paper derives the fundamental statistical theory which may serve as a basis for the planning of early detection clinical trials aimed at evaluating the benefit of early detection of disease combined with treatment. Early detection trials contain special characteristics which do not arise in therapeutic clinical trials. The lack of statistical theory for the planning of early detection trials has resulted in current trials being suboptimal. We develop probability models that are necessary for the planning of early detection clinical trials. Our models address four characteristics of early detection trials: (i) the optimal time of analysis and length of follow-up, (ii) the optimal spacing between examinations, (ii) the number of examinations versus sample size for fued cost and (iv) new experimental design where all participant groups receive an initial examination at the time they enter the study. Our optimization criterion is to maximize the power of the Application is made to breast cancer early statistical test for comparing mortalities. detection trials.
INTERNAL
Sll ESTIMATION OF THE SAMPLE SIZE FOR A t-TEST Jonathan S. Denne
Universityof Bath Bath, United Kingdom If the sample size for a t-test is calculated on the basis of a prior estimate of the variance, then the power of the test, at a treatment effect of interest, is not robust to misI propose a new t-test based on Stein’s two-stage specification of the variance. procedure, which uses an initial sample to estimate the variance and thus the fmal sample size required. This initial sample forms an integral part of the trial. My procedure controls the Type I and II error rates more closely than existing methods for the same problem. Furthermore, I extend my procedure to incorporate interim analyses. These both enable the trial to be stopped early if a large treatment diierence is observed, and permit the total sample size to be reestimated using the current estimate of the variance. Flexibility in the timing of interim analyses is achieved through the use of error spending functions. There are no existing methods detailed in the literature for dealing with this problem. Thii procedure also closely controls the Type I and II error rates. Keys words: Monitoring.