Row-column designs for investigational products vs. control comparisons in veterinary trials

Abstract

Veterinary trials are generally conducted for drug testing which uses experimental designs like parallel group designs, incomplete block designs, row-column designs, crossover designs, etc. In trials that are specially planned for making comparisons between investigational products with control(s), row-column designs eliminating variation due to periods and animals are most suitable. The present study demonstrated 2 methods of constructing small and efficient rowcolumn designs for investigational products vs. control comparisons. The first method provides designs (Series-I) for number of investigational products < 20 along with variance of estimated elementary contrasts pertaining to various comparisons of interest. The limitation of these designs is that these designs cannot be obtained for comparing odd number of investigational products with one (two) control(s). The second method (Series-II) yields designs for comparing all odd prime number of investigational products with one(two) control(s) and fills almost all the gaps left by Series-I. Thus, both series together provide a wide range of small and efficient row-column designs suitable for comparing a group of investigational products with one (two) control(s). The canonical efficiency factor of both the series of designs in comparison to an orthogonal design having the same number of investigational products has also been calculated considering a nested model with observational units nested within experimental units. These designs are found to be highly efficient.