Nested designs in ruggedness testing

Nested designs were performed in order to execute a ruggedness test accordi ng to the United States Pharmacopeia definition for ruggedness, in which ma inly non-procedure related factors are examined. Several nested designs hav e been executed on a high performance liquid chromatography assay to determ ine tetracycline and related substances in bulk samples of tetracycline. Fa ctors such as different laboratories, analysts, instruments, columns, days and batches were examined. The interpretation methods described in the lite rature were found to cause problems. In these methods the variances of the examined factors are estimated from the calculated mean square values and f rom the equation for the expected mean squares. Very frequently, negative v ariance estimates were obtained. Their absolute values were found to be dep endent on the influence of the factor examined below it in the design, on t he examined response. Therefore an alternative interpretation method for ne sted designs, based on pooled variances, was proposed and found to be appro priate to use for ruggedness testing purposes. Both approaches, the one fro m the literature and the one proposed here, were tested on simulated data c oming from a nested design with four factors and on the experimentally meas ured data. (C) 1999 Elsevier Science B.V. All rights reserved.