To the Editor:--The heavy statistical content and rapidly expanding terminology in studies with pharmacokinetic/dynamic modeling can make the significance of the results difficult to interpret. For instance, Kapila et al. report that the concept of context-sensitive half time (CSHTs) has been verified clinically because the measured CSHTs for two opioids are similar to their modeled CSHTs. Yet the similarities between the CSHT's shown in Table 3 are merely mathematical artifact of their methods. Their measured CSHT is obtained by fitting a single exponential curve to the plasma disposition data, and measuring the time for a 50% decrease in plasma concentration (the half life) after the 3-h infusion (the context). Their modeled CSHT is derived by fitting a multiexponential curve to the same plasma data, and then calculating the CSHT from the kinetic parameters that describe the curve. The number of terms in an Equation neededto describe a data set depends on the accuracy desired, so-called “statistical satisfaction”. For some data sets, the “goodness of fit,” measured objectively by a measure such as log likelihood, is not dramatically improved by adding more terms to the equation. The similarity between the measured and modeled CSHT in Table 3 is merely due to the single exponential Equation providinga “satisfactory” approximation to the multiexponential equation, but this has no clinical relevance. A more valid comparison for Table 3 would be the one made in the discussion, between the CSHT predicted by the model programmed into the infusion pump and the measured CSHT. This comparison is noteworthy because the utility of these studies to the clinician is the ability of these models to predict drug disposition and recovery without sampling drug levels. Making these comparisons in this study suggests, as have other recent studies, [3,4]that one can use kinetic sets from the literature to drive an infusion pump and have a “reasonable” expectation of drug disposition.
The authors finish by interpreting the pharmacodynamic offset of the drugs, a term they don't define. By inference, it's the time for the effect to decrease by 50%, as calculated from a single exponential curve fit to the data on recovery of minute ventilation. Confusion occurs when they use this term interchangeably with measured pharmacodynamic CSHT in Table 3. This seems inappropriate, because the pharmacodynamic CSHT, by definition, is the CSHT of the relation between the effect site concentration and the effect. This is usually a nonlinear, sigmoidal relation that includes measurement of maximum and minimum effects (Emax, E0) and cannot be defined by measurements made solely over the linear portion of the curve (40–70% decrease in ventilation), as was the study design in this case.
As we define the language of total intravenous anesthesia, we implore these and other authors laying the foundation to define and use their terms and statistical results with clarity and precision.
Frank J. Overdyk, M.S.E.E., MD; Assistant Professor, Raymond C. Roy, M.D., Ph.D.; Professor and Chairman, Department of Anesthesia and Perioperative Medicine, Medical University of South Carolina, Charleston, South Carolina 29425–2207
(Accepted for publication May 14, 1996.)