Predicted and measured plasma propofol concentration and bispectral index during deep sedation in patients with impaired left ventricular function.

Objective: To evaluate the ability of the Schnider pharmacokinetic model to predict plasma propofol concentration during target-controlled propofol infusion in patients with impaired left ventricular function and to investigate the predictive value of the bispectral index (BIS) to indicate deep sedation in this patient group.

Design: Prospective, observational study.

Participants: Thirty-four patients (mean left ventricular ejection fraction 31% +/- 9%) undergoing the implantation of a cardioverter-defibrillator during deep sedation.

Interventions: None.

Measurements And Main Results: Predicted and measured propofol plasma concentrations and BIS were assessed during steady-state conditions with the propofol infusion rate constant for at least 20 minutes. The plasma propofol concentration was significantly underestimated by the pharmacokinetic model used (mean percentage prediction error 37% +/- 49%). The 50% probability of deep sedation was calculated at a predicted propofol concentration of 2.09 (95% confidence interval [CI], 2.04-2.14) mug/mL and at a measured propofol concentration of 2.70 (95% CI, 2.62-2.78) mug/mL. BIS values showed a marked variability among individuals during deep sedation (5th-95th percentiles: 25-81).

Conclusions: The pharmacokinetic model used markedly underestimated propofol plasma levels in the patient group studied. The large variability among patients suggests that BIS monitoring is not suitable for indicating an exact endpoint corresponding to deep sedation.