Citation

Hubig M, Wehner HD. Blutalkohol 1997; 34(3): 183-205.

Affiliation

Institut fur Gerichtliche Medizin, Eberhard-Karls-Universitat, 72074 Tubingen, Germany

Copyright

(Copyright © 1997, International Committee on Alcohol, Drugs and Traffic Safety and Bund gegen Alkohol und Drogen im Straßenverkehr, Publisher Steintor Verlag)

DOI

unavailable

PMID

unavailable

Abstract

This paper presents a stochastic model for the probability distribution of blood alcohol concentrations (BAC) of randomly checked drivers. The model enabels us to reconstruct an empirical distribution P(EV) for the complete range of BAC levels if only the empirical distribution P(EK) of the higher BAC level is known. Such an incomplete BAC distribution is available, but due to police control regulations, is only valid for concentrations higher than the BAC c(lim). This model is based exclusively on parameters which can be interpreted physically and physiologically. It leads to a formula for a theoretically complete distribution P(TV) (referring to the complete range of BAC levels), of which a further formula for the theoretically incomplete distribution P(TK) (referring only to BAC levels higher than c(lim)) can be derived. An approximation of the latter distribution which accurately fits the empirical incomplete distribution P(EV) is calculated by optimising the parameters. Employing the parameter values in the formula of the theoretical distribution P(TV) valid for the complete concentration range and determined by means of the model, leads to the reconstruction of the probability distribution P(EV) of the complete range of the BAC of drivers. The present reconstruction procedure is a desideratum which makes it possible to use police traffic control data and accident reports in order to quantify the accident risk in drink driving. The presented model and reconstruction procedure is tested using data from the Grand Rapids Study (Borkesstein, 1964). In case of the complete empirical distribution P(EV) the distribution of control cases is assumed. In case of the incomplete empirical distribution p(EK) the distribution of the concentrations in the control collective restricted to levels higher than 0.75(per mille) is assumed. The adaptation of the theoretical distribution P(TK) and P(TV) to the respective empirical curves P(EK) and P(EV) is qualitatively good. The noticeable modulation of amplitudes in case of empirical curves P(EK) and P(EV) is reproduced by alcohol portions of 17.8 g and the double and triple of 7.7 g in the theoretical curves P(TK) and P(TV). The observed discrepancies between theoretical and empirical distributions can be explained by the small number of individuals on the one hand, and too little differentiation of the BAC levels on the other hand. Therefore an improvement of future results could be archieved by enlarging the collectives and by measuring the BAC more accurately. A further possibility of correction would be the simplification of the model. Given these results, an extension of the reconstruction procedure towards a reconstruction of the BAC distribution of accident drivers and a subsequent determination of the attributable risk seems very promising.

DWI