Estimating the power of a future clinical study is a common problem in the drug development process. Within the framework of model based drug development this problem is solved through Monte-Carlo studies where numerous replicates of the trial are simulated and subsequently analysed. This process can be very time consuming due to the high number of replicates required to obtain a stable power estimate. Non-linear mixed effect models which are frequently used for the analysis of clinical trial data are especially problematic as they can have a run time of several hours.
A novel parametric power estimation (PPE) algorithm utilizing the theoretical distribution of the alternative hypothesis is presented in this work and compared to classical Monte-Carlo studies. The PPE algorithm estimates the unknown non-centrality parameter in the theoretical distribution from a limited number of Monte-Carlo simulation and estimations. Furthermore, from the estimated parameter a complete power versus sample size curve can be obtained analytically without additional simulations. The PPE and classical Monte-Carlo algorithms were compared for 3 different drug development examples.
For a single power calculation, given a specific sample size, the PPE algorithm provided accurate estimates for all investigated scenarios and required 2 times fewer samples than the pure Monte-Carlo method to achieve the same level of precision. Furthermore, from this single power calculation, the PPE method can derive an entire power curve (power versus sample size), drastically reducing run times for this computation. The power curves from the PPE algorithm were in excellent agreement with the curves obtained using classical Monte-Carlo techniques.