Sedation for endoscopy is a rapidly emerging endeavor in anesthesia. Growth in this area has been steady, and anesthesiologists are increasingly becoming involved in endoscopic sedation. Endoscopy is distinguished from other anesthetic challenges by 2 factors. First, these procedures are performed with natural airways, and excessive sedation may induce obstruction and respiratory depression. Second, the procedure time is short, and there is insufficient time to tune the anesthetic. These factors affect the anesthetic strategy. Anesthesiologists assume that the skills learned in the operating room transfer to the endoscopy suite, but a bolus of propofol sufficient for a 99% probability of obtunding response to intubation may exceed the total propofol requirement for a diagnostic esophagogastroduodenoscopy (EGD) several times and result in a prolonged period of jaw thrust to overcome obstruction. Conversely, starting a propofol infusion at the infusion rate for maintenance of loss of consciousness will take a considerable period of time to achieve this outcome. Target-controlled infusion (TCI) may achieve a specified effect-site concentration reliably, but the variability of patient response complicates the selection of the target.1 Thus, in a relatively short encounter, anesthesiologists must pick the appropriate induction dose for deep sedation and from this infer the proper maintenance dose.
Experienced clinicians use a number of strategies, but these can be divided into 2 broad camps based on whether an infusion pump or a handheld syringe is employed in titration. This chapter will present 2 strategies that use a pharmacokinetic model of propofol with little more than a few bits of information, a watch, and a calculator. This model enables anesthesiologists to optimize sedation and analgesia for EGD procedures that rivals the performance of real-time optimal control algorithms. It is important to recognize that these techniques are not a substitute for years of experience or sophisticated monitoring technology. Rather, they represent an application of drug simulation at the point of care that permits a novice to consistently “hit the dartboard.”
Simulations will be used to illustrate the 2 dosing strategies. These simulations will utilize a propofol 3-compartment pharmacokinetic model introduced by Cortinez et al.2 This model permits consideration of increasing weight without the problems associated with the high body mass indices encountered in earlier models of propofol pharmacokinetics. The model parameters are included in Table 34–1.
Compartment volumes are scaled by weight/70 kg, and clearances by weight/70 kg raised to the 0.75 power (referred to as allometric scaling). Age corrections are applied to model parameters V2 and Q2, as indicated in Equations 34–1 and 34–2.
where TBW is total body weight. An effect-site is adjoined, with ke0 calculated to yield a time to peak effect of 1.6 minutes. Simulations will also use a pharmacodynamic model of loss of consciousness based on estimated propofol effect-site concentrations.3 Unless otherwise stated, simulations throughout this chapter are for a 50-year-old, 70-kg patient undergoing an EGD.
In developing strategies to rapidly determine the appropriate depth of anesthesia for a given patient, several assumptions will be used:
A loading sequence (ie, a slow bolus or rapid propofol infusion) can be determined that will cause a smooth increment in the probability of achieving loss of consciousness that is similar across all patients.
If the rate of change in probability of loss of consciousness is low, it is possible to infer the patient’s sensitivity from the time to loss of consciousness.
Given this estimate of sensitivity, a maintenance sequence (ie, either intermittent boluses or a continuous infusion) will be identified that will maintain the effect-site concentration associated with loss of consciousness for the short duration of the procedure.
For the purposes of this effort, we will define loss of consciousness as lack of response to verbal stimulus.
For each strategy, a control system is used to find the optimum solution to the titration sequence. This is done by defining a desired trajectory for the effect-site, as described in Technique One. The infusion sequence is then adjusted so that the predicted trajectory is as close to the specified trajectory as possible. This is done by repeatedly running the simulation with adjustments in the infusion sequence until the difference between the desired and predicted trajectories (termed the error) is as low as possible, perhaps even zero. This process requires seconds on a modern computer running specialized software, such as MATLAB. Although this is no obstacle to the author, to make the examples workable by the reader, the strategies have been restated as recipes that can be “scaled to the size of the dinner party.”
The first strategy is a bolus technique suitable for a handheld syringe or a pump used in bolus mode (a patient-controlled anesthesia [PCA] pump such as the Graseby 3300 works well for this). It begins with an initial loading dose of propofol, typically 20 to 80 mg, followed by small fixed doses (one-fifth the loading dose) until adequate sedation is obtained, which is then repeated periodically to maintain adequate sedation. The advantage of this technique is that it can be performed with a schedule for boluses. A combined pharmacokinetic–pharmacodynamic model can be used to advise clinicians when to administer the boluses, as was previously reported.4 Model predictions of effect-site concentrations, drug effect (in this case unresponsiveness), and suggested bolus dosing regimen are illustrated in Figure 34–1.
Propofol bolus dosing regimen (top plot), resultant predictions of propofol effect-site concentration (Ce) levels (middle plot), and probability of unresponsiveness (bottom plot) using intermittent small propofol boluses. The intermittent boluses are presented as burst from an infusion pump that delivers propofol at a rate of 3 mg/sec (1 mL every 3 seconds). The first bolus is for 10 seconds and subsequent boluses are for 2 seconds. The duration of time between boluses is progressively shorter (top plot).
An initial bolus of 30 mg is administered over 10 seconds; subsequent 6-mg boluses are administered over 2 seconds. The interval between boluses decreases as a function of time, and the depth of anesthesia can be controlled by the period between boluses. To implement this strategy, all that is needed is a value for the loading dose and the schedule for bolusing. This dosing regimen can be implemented by following these steps:
Administer the loading dose.
Wait 72 seconds.
Administer the first incremental dose.
Wait 36 seconds.
For every subsequent incremental dose, decrease the waiting time by 5 seconds (to a minimum of 2 seconds).
Repeat until the patient loses consciousness up to a total of 20 incremental doses.
For the loading dose and 20 incremental doses, a total of 161 mg (2.3 mg/kg) is required for a 99% probability of loss of responsiveness within 5 minutes. There will be a few patients for whom this will be inadequate, but most patients will require less, and the average patient should be unresponsive by 3 minutes and 50 mg.
Once the dosing sequence to achieve loss of unconsciousness is identified, the bolus sequence that will maintain that effect-site concentration can be determined, as shown in Figure 34–2. With this technique, there is a slight overshoot because there was no way of knowing when a patient will lose consciousness—in this example, shortly after the third bolus. The bolus rate, however, rapidly stabilizes at a constant interval for the duration of the brief anesthetic, and is well approximated by a linear function of the time to loss of consciousness (T).
For maintenance, 3 more steps are added to the previous list.
When the patient loses consciousness, note the elapsed time T (in seconds) from start of the loading dose.
The bolus interval I is 78 – 0.1825 × T (seconds).
After loss of consciousness, give the next bolus 1.25 × I seconds after the preceding bolus.
Propofol bolus dosing regimen to achieve and maintain unresponsiveness in a patient of average sensitivity. The top plot presents the bolus dosing regimen, the middle plot presents the predicted propofol effect-site concentration (Ce), and the bottom plot presents the predicted probability of unresponsiveness. In this example, the patient looses responsiveness at a 50% probability of unresponsiveness (ie, in a group of 100 patients, with this exact dosing regimen, 50 would be unresponsive and 50 would be responsive).
To use this technique in clinical practice, a PCA pump can be programmed to deliver the appropriate bolus, and a computer can be programmed to beep every time a button press is needed. Although this method was not easy, it was successfully used to manage 25 patients undergoing colonoscopy.4 It provides an appreciation of what can be accomplished using application of propofol pharmacokinetics to automatic control.