Advances in quality control for dioxins monitoring and evaluation of measurement uncertainty from quality control data.

This paper describes an application of multivariate and multilevel quality control charts with the aim of improving the internal quality control (IQC) procedures for the monitoring of dioxins and dioxin-like PCBs analysis in food. Dioxin analysts have to use the toxic equivalent concept (TEQ) to assess the toxicity potential of a mixture of dioxin-like compounds. The TEQ approach requires quantifying individually 29 dioxin-like compounds. Monitoring the congeners separately on univariate QC charts is misleading owing to the increase of false alarm rate. We propose to subdivide the TEQ value into 3 sub-groups and to control simultaneously the 3 variables in a T(2) chart. When a T(2) exceeds the upper control limit, it acts as a warning to trigger additional investigations on individual congeners. We discuss the minimum number of runs required to reliably estimate the QC chart parameters and we suggest using data from multilevel QC charts to properly characterize the standard deviations and the correlation coefficients. Moreover, the univariate QC chart can be sensitised to detect systematic errors by using exponentially weighted moving average (EWMA) technique. The EWMA chart provides an additional guidance on setting appropriate criteria to control the method bias and to support trend analysis. Finally, we present an estimate of measurement uncertainty by computing the accuracy profile in a retrospective way with the QC data generated and we discuss assessment of compliance with regulatory maximum levels.