It may be beneficial to re-estimate the sample size of clinical trials based on blinded assessment of interim data. For example, in studies using a discrete outcome (event) endpoint, if the observed overall study event rate is lower than the expected rate, one may increase the sample size without affecting the type I error rate since interim data is blinded and treatment difference is not assessed. However, if the observed overall event rate is too low compared to the expected rate, it sometimes makes both scientific and commercial sense to terminate the study. Such practices prevent running costly trials with little chance of success without affecting the overall type I error rate of the study.

Questions: Do we need to prespecify blinded interim assessment in the protocol or SAP? Is it beneficial to include cost into the consideration of the sample size adjustment in addition to the statistical aspects of the study? Is it necessary to conduct a risk/benefit assessment analysis that includes both the cost and the chance of success?