Abstract:
|
There are many decision points during the clinical development of a new drug. These decisions have traditionally been based on the hypothesis-testing framework where type-I error and power requirements drive the sample size calculation. In this course, we will take a different approach by treating clinical trials as a series of diagnostic tests in which the goal is to estimate the likelihood that a drug has the desired profile. Treating a trial as a diagnostic test translates the concept of power and type-I error rate of the former to the sensitivity and 1-specificity of the latter. Positive predictive value now refers to the probability that a new drug has the desired properties. This analogy facilitates formal incorporation of evidence from a previous trial into the design of a future trial and the subsequent decision criteria, allowing the formulation of go/no-go criteria that can address the unique needs of different stages of the clinical testing. Using the above analogy, we will discuss different metrics relevant to decision making at the proof-of-concept, dose-response, and confirmatory stages. We show how appropriate metrics may enable better decisions and illustrate several potential mistakes trialists should guard against. Examples will be offered throughout the course.
|