Sources of Random Variation of Pesticide Residue Analytical Results.

Background: Pesticide residues are analyzed in thousands of samples yearly by national authorities and private laboratories. Intensive research is ongoing to develop new methods or improve existing ones concentrating on extraction, cleanup, and detection techniques. Little attention has been paid to the contribution of prior steps in the determination process to overall laboratory sampling errors, though several publications demonstrate their practical importance. Consequently, the repeatability and reproducibility of the results are often reported based on the recovery tests alone. A few previous publications are cited in this paper which illustrate the magnitude of random errors derived from subsampling, comminution of analytical samples, and selection of small test portions. Objectives: We aim to call attention to the importance of considering all steps of laboratory sampling and analysis processes in calculating the combined uncertainty of results and realistic performance assessments of methods including their longterm intermediate precision. Method: Validation of laboratory sampling of large fruits is used to illustrate the recommended procedures, determination of their random error, and long-term method performance. Results: The results indicate that subsampling, comminution, and selection of test portions can be major contributors to the combined uncertainty of results. Conclusions: All these steps should be considered in estimation of random variation (uncertainty) of measured residues. Highlights: Random error of laboratory sampling for pesticide residues. Mass reduction of large crop units. Internal quality control of laboratory operations. [ABSTRACT FROM AUTHOR]