Assessment of measurement uncertainty using longitudinal calibration data in the forensic context

Abstract

Suppose that results of the scientific literature were presented in courts the same way they are presented in publications. In the present investigation, it is assumed that this is the case, and it is shown that it may potentially lead to issues with results that are being presented to the court. The determination of synthetic cannabinoids at low concentrations in blood, urine and other matrices is challenging to forensic science. Methods based upon the separation of the compounds by HPLC and detection by tandem mass spectrometry provides results at ultra-low concentrations. The uncertainty of measurements has become the key parameter of interest for decision making, and expert witnesses need to state the correct level of uncertainty. Recent developments in quality assurance indicate issues with reliability, owing to contradictory statements that originate from different methods of data management and interpretation of results that eventually may lead to compromising the truth. The levels of relative uncertainty of measurement that were close to 5% were found to be unrealistically low when synthetic cannabinoids were analysed at ultra-low concentrations. It was proposed to introduce the principle of pooled calibrations to obtain correspondence between predicted and observed uncertainty, following standards of scientific methodology. Ten synthetic cannabinoids were analysed with pooled calibrations, and the results indicated that the uncertainty of measurement was found at levels much higher than expected, and with two out of ten synthetic cannabinoids that were impossible to quantify with relative uncertainties reaching levels over 70%.