Pharmaceutical Statistics Using SAS

14.5. Optimal Sample Size

The Phase III investment example resembles investment problems from other industries, e.g., oil drilling examples described in Raiffa and Schlaifer (1961, Section 1.4.3). One important difference is that Phase III development consists of experiments which we are to design and whose results can be modelled with standard statistical methods. The power, or probability of showing superiority, in one clinical trial is easily determined given the response distributions and test method. The response distributions are typically determined by parameters such as mean effect and standard deviation. Furthermore, these parameters can be modelled using previous clinical (and sometimes preclinical) data (Pallay and Berry, 1999; Burman ...