Institute of Legal Medicine and Traffic Medicine, University of Heidelberg, D - 69115 Heidelberg, Voßstraße 2, Germany

According to the German standard DIN 32645 (DIN = Deutsches Institut für Normung) the limit of detection of an analytical method is statistically calculated from calibration data. First, the "critical value" is determined at which the reading of a sample exceeds that of a blank sample taking a normal ("Gaussian") frequency distribution of zero readings and a probability of error (a = 0.01) into account. Depending on the critical value, the limit of detection (LOD) is defined as a 50% probability of classification of the analyte as present (a = 0.5) or not present (b = 0.5). The limit of determination (LODn) is the smallest quantity which can be detected with a probability of 1-b (b = 0.05 or 0.01). The limit of quantitation (LOQ) refers to the minimum quantity which can be determined with both defined probability level (b < 0.01) and acceptable relative uncertainty.

We calculated these limits for THC in serum using a computer program, our calibration data and three specific ions [m/z: 371 (100), 386 (93), 303 (60)]. For forensic issues and previously unknown presence of THC, neither LOQ nor LODn may describe the identification performance properly if the most intensive ion (quantifier) is used for calculation. Calculated from the more intensive ions, the LOQ _THC was = 1.52 µg/L serum. The Limit of Identification (LOI) characterizing a GCMS-method's identification power was obtained using the least intensive ion (m/z 303) for the calculation of the LODn. A LOI_THC of about 2.7 µg/L serum indicates that below this concentration THC will be detected and can be quantified, however, as it may not allways be identified. Its identification must then be specified.

Our data show that these terms are related to each other, to concentrations used for calibration in a single matrix and to the coefficient of determination of the individual regression line. However, in individual matrices (blood samples) the concentration at which identification or quantitation is possible may deviate from these theoretical data. In contrast, analytical sensitivity described by an analytical threshold value would mean a definite concentration at which the analyte will be identified with a probability of detection of 0.95 regardless of the property of the individual serum matrix.