Predictive performance of target controlled infusion of propofol-MCT/LCT using the modified Marsh and Schnider models: a simulation study

ABSTRACT

BACKGROUND: Only two pharmacokinetic models of propofol are commercially available in the category of target controlled infusion (TCI) pumps the modified Marsh and Schnider models. Both models were developed using propofol-LCT (long chain triglyceride). Depending on the excipient, the pharmacokinetic properties of fast-acting drugs, such as propofol, vary. Hence, it is necessary to evaluate the predictive performances of both models using propofol-MCT (medium chain triglyceride)/LCT, which is frequently used in clinical practice. METHODS: This was a computer simulation study, using data collected in the previous clinical analysis used to evaluate the predictive performance of a pharmacokinetic model of propofol-MCT/LCT. The infusion profiles for each patient were applied as inputs to both models. Simulations were performed using TCI software, and the simulated plasma concentrations of both models were calculated. RESULTS: In total, 217 plasma samples, obtained from 35 patients, were used to determine the predictive performance. The pooled median (95% CI) biases and inaccuracies were 9.6 (−1.7 to 15.4) and 32.1 (22.6–38.2) respectively, for the modified Marsh model, and −5.9 (−8.9 to −0.7) and 26.3 (21.7–27.8) respectively, for the Schnider model. CONCLUSIONS: Although the pooled bias and inaccuracy of the Schnider models were clinically acceptable (< 10–20% and approximately 20–30%, respectively), the Schnider model consistently produced negatively biased predictions. Conversely, even though the pooled inaccuracy of the modified Marsh model failed to meet this criterion, the value did not deviate significantly from the standard. Therefore, it is reasonable to conclude that both TCI models can be used for propofol-MCT/LCT.