Small sample approach, and statistical and epidemiological aspects.

In this chapter, the design of pharmacokinetic studies and phase III trials in children is discussed. Classical approaches and relatively novel approaches, which may be more useful in the context of drug research in children, are discussed. The burden of repeated blood sampling in pediatric pharmacokinetic studies may be overcome by the population pharmacokinetics approach using nonlinear mixed effect modeling as the statistical solution to sparse data. Indications and contraindications for phase III trials are discussed: only when there is true "equipoise" in the medical scientific community, it is ethical to conduct a randomized clinical trial. The many reasons why a pediatric trial may fail are illustrated with examples. Inadequate sample sizes lead to inconclusive results. Twelve classical strategies to minimize sample sizes are discussed followed by an introduction to group sequential design, boundaries design, and adaptive design. The evidence that these designs reduce sample sized between 35 and 70% is reviewed. The advantages and disadvantages of the different approaches are highlighted to give the reader a broad idea of the design types that can be considered. Finally, working with DMCs during the conduct of trials is introduced. The evidence regarding DMC activities, interim analysis results, and early termination of pediatric trials is presented. So far reporting is incomplete and heterogeneous, and users of trial reports may be misled by the results. A proposal for a checklist for the reporting of DMC issues, interim analyses, and early stopping is presented.