Probability Ramp Control of Propofol for EGD

Administration of propofol to achieve a target of moderate sedation is a challenging task for which anesthesia providers receive minimal training. Undersedation results in a noncompliant patient, while oversedation results in airway obstruction, respiratory depression, and hypotension. Considerable variability in patient pharmacokinetics (the distribution of drug within the body) and pharmacodynamics (the translation of drug concentration to clinical effect) has been demonstrated. The skill of titrating propofol to the desired target and maintaining this state is slowly acquired in the clinical environment of the endoscopy center with frequent reliance on rescue skills. An automated system that facilitates this process would be useful.

Pharmacokinetic models allow us to make predictions of the results of drug administration. If we know the age and size of the patient, we can determine a quantity of propofol that will attain a desired concentration at some point in the future (within the predictive accuracy of the model). If they are old, this is less than if they are young. If they are obese, this is more than if they are thin. By adjusting the dosing, we can achieve similar concentrations at a specified time in a wide range of patients.

Pharmacodynamic models allow us to relate drug concentration to a probability of response. Sensitivity is a randomly distributed variable, and the cumulative probability of response to propofol is well represented by a sigmoid curve. While we do not know the concentration that will suffice for a given individual, we can determine the probability that this individual will lose responsiveness within an interval of concentrations. For example, the probability of loss of responsiveness between 1 µg/ml and 6 µg/ml is around 99%. For any given age and size, an infusion sequence can be determined so that we traverse this interval smoothly. The infusion sequence is determined by minimization of the difference between the simulated probability and the target (1). We predict that 90% of 50 year old 70 kg patients will lose responsiveness between one minute and three minutes after initiating the infusion, and 99% by five minutes. The infusion sequence for this patient is comprised of a bolus of 287 µg/kg followed by an initial infusion of 216 µg/kg/min, with an increase to 550 µg/kg/min after 147 seconds. By selecting the infusion sequence based on the age and size of the patient, all patients will track the same target line. These infusion rates are determined prior to initiation of sedation, and the clinician can verify that they are appropriate for the patient before beginning sedation.

Once the endpoint of adequate sedation is observed, the effect site concentration associated with this endpoint is inferred, and the infusion that will maintain this concentration can be determined. This allows the clinical observation to be translated into an infusion rate, much as a driver accelerates to a desired speed and then engages the cruise control to maintain that speed.

The intent of this study is to demonstrate equivalent safety and efficacy of PRC to control by a skilled clinician.

References

1. Mandel JE, Sarraf E. The Variability of Response to Propofol Is Reduced When a Clinical Observation Is Incorporated in the Control: A Simulation Study. Anesthesia \& Analgesia. 2012;114:1221-9.