NONCLASSICAL EXPERIMENTAL-DESIGNS IN PHARMACEUTICAL FORMULATION

Classical experimental designs suitable for studying the effect of the proportions of the components on the properties of a mixture are readily developed and are of optimum efficiency. Their applicability depends however on the experimental factor space having a very simple form, and because of the many limits imposed on the levels of pharmaceutical excipients in pharmaceutical dosage forms their practical use in formulation is restricted. Methods are available for defining the experimental factor space where it is not a simplex and the number of components is too large for the space to be defined graphically. A relatively easy method of obtaining good experimental designs is to use the D-optimum criterion, where a given number of experiments is selected out of the many possible mixtures, to give a statistically optimized design. An example of the use of this method in formulating a modified release matrix tablet, containing 5 variable components is described. One of the advantages of the method is the possibility of a step-wise approach, and this is demonstrated in the present case where the formulation was carried out in two stages. The results enabled the formulation of tablets with the desired dissolution characteristics together with a fairly complete characterisation of the system.