Problems in Pharmacokinetic Analysis of Alcohol Disposition: A Trial of the Bayesian Least‐Squares Method

Abstract

The technical problems of the pharmacokinetic analysis of alcohol disposition were studied using the Michaelis-Menten elimination kinetic model. This model was defined by two forms of equations: differential and integrated, with the latter being derived by integration of the differential equation. We compared the parameter values, estimated by one-line curve-fitting, using these two equation forms. We concluded that, for the kinetic analysis of alcohol disposition, curve-fitting with the differential equation was superior to that with the integrated equation. We also studied the methodological problems involved in one-line fitting. The ordinary least-squares (OLS) method was compared with the Bayesian least-squares (BLS) method. Correlation between the Vmax and beta (ethanol elimination rate) values, and between the Vmax and K(m) values was seen when the parameter values were estimated by the OLS method. These results suggested that one-line fitting by the OLS method was not adequate for Michaelis-Menten-type elimination kinetic analysis. BLS analysis resulted in no correlation between the estimated parameter values that did not change with the level of the dose. The BLS method seemed to be more useful than the OLS method for the estimation of individual pharmacokinetic parameter values.