sasによる 中間解析のデザインと解析 - sascom.jp.pdf

SAS Interim Analysis using SAS Chikuma Hamada Tokyo University of Science V9.2 SEQDESIGN SEQTEST SEQDESIGN,SEQTEST, 2 1 O‘Brien-FlemingPocock SEQDESIGN SEQT EST 3 αβσΔN Δ N σ I α β 4 2 1)You specify the statistical details of the design, including the null and alternative hypotheses, a test statistic , the Type I and II error probability levels, a stopping criterion, the total number of stages, and the relative information level at each stage. 2)You compute the boundary values for the trial based on the design specifications. T he sample size required at each stage is also computed for the specified hypothesis test. 3)At each stage, additional data with the required sample sizes are collected. T he data available at each stage include the data collected at the current stage as well as the data collected at previous stages. 4)At each stage, the available data are analyzed with a procedure. 5)At each stage, you compare the test statistic with the corresponding boundary values. Either the trial is stopped to rej ect or accept the hypothesis, or the trial is continued to the next stage. If the trial continues to the final stage, the hypothesis is either rej ected or accepted. 6)After the trial stops, you compute parameter estimates, confidence limits for t