Applications of Probability of Study Success in Clinical Drug Development

Abstract

The dominant approach to sample size determination for a clinical trial in regulatory review-driven pharmaceutical research has long been by assuming fixed values of parameters under competing hypotheses, i.e., null versus alternative representing futility and desired efficaciousness of a tested drug. A sample size is then determined to ensure sufficient statistical power for differentiating between the null and alternative hypotheses, while controlling the probability of wrongly rejecting the null. This approach bears the criticism of ignoring the variability inherent with the unknown parameters. To improve sample size determination, accounting for variability of parameters has recently been gaining application in pharmaceutical-conducted clinical trials. The common intent of this increased interest is to better predict the probability of a successful trial, which is often termed probability of study success (PrSS) or probability of success (POS). We discuss the important role that PrSS can play in clinical trial design and decision making throughout medical product development. A few examples are given for illustration.