Sample Size Re-Estimation with Missing Data in Clinical Trials

• Activity Number: 534
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Ruitao Lin*+ and Guosheng Yin and Huiqiong Li
• Companies: Department of Statistics and Actuarial Science, University of Hong Kong and University of Hong Kong and Yunnan University
• Keywords: Clinical trials ; Adaptive Design ; Sample size ; Missing Data ; Multiple imputation ; Conditional power

Abstract

In clinical trials, patients may drop out with various reasons, such as treatment failure, excessive toxicity. Such missing data may cause efficiency loss and thus make the trial underpowered, some may even induce biased estimations and incorrect statistical inferences. Often, sample size calculation does not incorporate missing data formally at the trial design stage, because the amount of missing data and the missing pattern are typically unknown in advance. Therefore, it is natural to take an interim analysis during a trial to evaluate the impact due to missing data thus far. Under the mechanism of missing at random, we propose to utilize multiple imputation to re-estimate the sample size at the interim stage of a clinical trial. Based on the conditional power approach, we can adaptively update the sample size for the rest of the trial in order to achieve the desired power. In addition, the proposed method can be implemented to deal with incorrect sample size estimation due to the misspecified effect size or variance. Extensive simulation studies are conducted to investigate the performance of our method, and a real trial example is used for illustration.