Design of a quality control scheme to assess sample preparation performance for the determination of deoxynivalenol in wheat

While proficiency testing is a useful tool to assess and monitor the performance of an analytical method, the use of comminuted test samples precludes the assessment of sample handling and preparation. These stages of the measurement process can introduce bias and significant variance into testing results. In this work, two approaches were used to prepare test material consisting of whole grain wheat containing a known amount of deoxynivalenol (DON) to be used to assess the variance due to sample preparation. The successful approach produced wheat kernel-like material from dough made with an aqueous DON solution. The produced material was physically similar to wheat kernels, with realistic DON content (mean 633 mg/kg) and low kernel-to-kernel variation (5% relative standard deviation). Test samples of whole grain durum wheat were prepared to approximate real-world samples with 0.2% fusarium damage. Fourteen participants analysed the whole grain test samples using their own sample preparation and analytical test methods. Calculated sample preparation variance, influenced by sub-sampling and comminution of test samples, varied from 0.0043 to 1.704 mg2/kg2. Sample preparation variance was significantly lower for participants that had comminuted the entire test sample as opposed to comminuting only a portion. The test samples produced provided participants with a straightforward and controlled process to assess their sample preparation and whether it is fit for their purpose.