Performance criteria for rapid screening methods to detect mycotoxins

Abstract

The paper describes the validation of screening methods that are used for official control to classify samples into negative and suspect positive samples. The concept is based on the principle that negative samples are considered as compliant, whereas suspect positive samples need to be re-analysed with confirmatory methods. An important performance criterion often used is a maximum value of 5% for the probability of false negative results obtained on samples that contain the analyte at the legal limit. Since the result of analysis is a binary decision, specific validation schemes need to be applied. The paper places emphasis on practical aspects of the calculation of the method performance characteristics, which are required to check whether the methods fulfil the performance criterion. The paper shows that screening methods based on a visual inspection, e.g. a dipstick, require special data treatment. In contrast there are many methods where the classification into negative and suspect positive samples is based on the comparison of a measured response against a cut-off value. This type of methods can be validated with quantitative statistics. The paper also elaborates on the calculation of the rate of false positive results of compliant samples. In addition the impact on the economical aspect of the use of the screening method is estimated, taking into account external factors such as the cost ratio between the screening and the confirmatory method and the occurrence of non-compliant samples in the entire population of the samples.