Incerteza de medição em análises micotoxicológicas: estimativa pelas abordagens Bottom up, Monte Carlo e Kragten

Abstract

Different approaches for the estimation of the uncertainty related to measurement results are found in the literature and in published guidelines. In the present work three approaches were used to estimate the uncertainty of the determination of aflatoxins (AB1, AB2, AG1, AG2) and deoxynivalenol in maize by liquid chromatography coupled to mass spectrometry tandem (LC-MS/MS): the Bottom up approach, adapted from the Guide to the Expression of Uncertainty in Measurement (ISO GUM); the Monte Carlo Method (MCM), which propagates distributions assigned to the input quantities through a numerical simulation; and the Kragten approach wich calculates standard deviations and confidence intervals with a universally applicable spreadsheet technique. A measurement equation was developed for mycotoxins analysis and a cause-and-effect diagram was draft to assist in the identification of the sources of uncertainty associated with the method. Detailed analysis of contributions of the various uncertainty sources was carried out. Measurement uncertainty was determined by the addition of the variances of the individual steps of the test procedure, according to each approach employed. The Bottom up, MCM and Kragten approaches produced very similar estimates to the combined uncertainty and the coefficient of variation (CV) between them was smaller than 1.0%. The main contribution to overall uncertainty is the intermediate precision with contributions over 90,0% for each mycotoxin. The results obtained with this research conclude that the three approaches are adequate for estimating the uncertainty in mycotoxin assays with LC-MS/MS technique. Among them, Bottom up is the most appropriate approach, since it requires that the analyst performs a detailed investigation about dominant components of the measurement uncertainty, allowing for better understanding and improvement of the measurement process. The Monte Carlo and Kragten methods are indicated the data generated by the Bottom up approach.