ABSTRACT:
The analysis of volatile congeners of alcoholic beverages is regularly used as an efficient tool for examining claims of drinking after committed offences. In contrast to the relatively simple Widmark formula for calculation of the ethanol concentration backwards to the offence time, the concentrations for every single volatile congener have to be calculated using complex exponential type formulas. To examine the claims of alcohol consumption, the calculated values of the volatile congeners are usually compared to data from the literature. The aim of this study was to evaluate if multivariate statistic techniques provide a simpler means of checking claims of drinking.
In a drinking trial, five participants consumed an amount of the same brand of German brandy calculated to reach 0.8‰ of blood ethanol concentration. Venous blood was taken at 30 min, 90 min, and 150 min after the drinks. Urine was taken four times after the drinks. In a second trial, the same participants consumed a second brand of German brandy under the same conditions. The alcoholic beverages, blood and urine samples were analysed using validated standard gas chromatographic methods. Then Principal Component Analysis (PCA) was used to transform the original measurement variables of the alcoholic congeners into new variables called principal com-ponents (PC). By plotting the data in a coordinate system defined by the two largest principal com-ponents, it was possible to identify key relationships in the data as well as to find similarities and differences.
Using PCA, outlying samples of one participant were identified, which could be attributed to unstated drinking of other alcoholic beverages in non-compliance with the experimental design. A classification between the two brands of brandy was possible in the blood taken at 30 min and the urine taken in a short time interval after drinking. In conclusion, the determination of the brand of consumed spirit is possible if samples are obtained shortly after drinking.
Besides, it was found during the trial that the expected values calculated after Bonte’s formula gave a better correlation to the analysed values if current analysis results of the spirit drinks were rather used than data from older standard references.
