Assessment of the clinical validity of an adjusted Marsh pharmacokinetic model using an effect-site rate constant (k_e0) of 1.21 min^-1

Abstract

Introduction: The first commercially available target-controlled infusion pump, the “Diprifusor”, employed a “Marsh” pharmacokinetic-pharmacodynamic parameter set, and this model (the Diprifusor-Marsh model) is also available in several of today’s pumps. However, it is unsuited for effect-site, target-controlled infusions (C_e-TCI) because it assumes slow transfer between blood and the effect-site (k_e0 = 0.26 min^-1). We hypothesised that a faster ke0 of 1.21 min^-1 (Adjusted-Marsh model) for C_e-TCI would result in hypnotic effects equivalent to that of the Schnider parameter set (Schnider model).

Methods: We replicated a previously published study that demonstrated the Diprifusor-Marsh model’s unsuitability for C_e-TCI. We randomised 40 unpremedicated young adults into two groups to receive C_e-TCI, employing either the Schnider model or the Adjusted-Marsh model. We infused propofol at 3 000 mg.hour^-1, while running a pharmacokinetic simulation and recording the electroencephalographic bispectral index (BIS) electronically. At loss of consciousness (LOC), indicated by a syringe-drop, we converted the infusion to C_e-TCI, targeting the effect-site concentration (C_e) observed at LOC, for 20 minutes. We regarded a difference of 10 BIS-units as clinically important.

Results: There were no statistically significant differences between the group medians regarding time-to-LOC, induction-dose, BIS at LOC and C_e-target. BIS decreased monotonically in both groups from a median of 78.5 at LOC to a steady-state median (25–27) at 15 minutes. The BIS of the Adjusted-Marsh model group closely followed the BIS of the Schnider model group. At steady state, the median BIS difference (95% CI) was -0.3 (-5.7 to 5.3), which was within the predefined interval for declaring equivalence. The Schnider model group’s mean BIS at steady state did not differ from that of the previous study’s Schnider model group.

Conclusion: Reasons for the progressive BIS decrease to lower than expected values include delayed response-times by the BIS monitor, C_e overshoot explained by front-end kinetics, neural inertia and the choice of surrogate LOC indicator. We conclude that C_e-TCI using the Adjusted-Marsh parameter set results in equivalent hypnotic effects to those of the Schnider model, with the proviso that this may apply only to young adults of normal body habitus.