Evaluation of sample processing and sub-sampling performance Original Research Article

A methodology for the estimation of sample processing and sub-sampling performance based on the comparison of the global method experimental dispersion of results with the uncertainty estimated from developed models for the subsequent analytical steps is presented. This approach is a valuable alternative to the evaluation of adequate experimental information using a classical ANOVA, since the significance of the sample processing and sub-sampling is evaluated with a higher number of degrees of freedom for the same number of experimental assays, due to the high number of degrees of freedom associated with the uncertainty estimated for the subsequent analytical steps from the combination of the involved sources of uncertainty. Considering the construction of a model to describe the performance of the analytical steps following sample processing and sub-sampling over a broad concentration range, the experimental assays involved at the evaluation of the sample processing and sub-sampling can be performed at any concentration meeting the previously validated range and several months after the development of that model once its adequacy has been proven over time. This approach, which also allows the construction of a detailed performance model for the global analytical method over a broad concentration range, was applied to the determination of pesticide residues in apples by gas-chromatography with electron capture detector. Considering that no information was available regarding the samples heterogeneity, sub-sampling performance was evaluated considering a sample representing the worst expectable homogeneity. This was accomplished by spiking just one out of the 10 halves of apples processed in each sample. The developed model for the performance of the analytical method was successfully and easily applied to routine analysis through an automated link between the information generated by the chromatograph software with a file containing the model.