Chapter 40. Principles of Pharmacokinetics and Pharmacodynamics: Applied Clinical Pharmacology for the Practitioner

1. The ultimate goal of pharmacokinetic–pharmacodynamic study is the accurate prediction of the time course and magnitude of drug effect so the clinician can answer a very simple and important question: "What is the appropriate dosing scheme for my patient?"

2. Pharmacokinetics, often thought of as "what the body does to the drug," is the study of the relationship between the drug dose and the drug concentrations that are produced over time.

3. Pharmacodynamics, often thought of as "what the drug does to the body," is the study of the relationship between the drug concentration and the drug effects that are produced.

4. Pharmacokinetic–pharmacodynamic models can be constructed that characterize drug behavior. These models are mathematical expressions of the relationship between drug dose and concentration (pharmacokinetics) and drug concentration and effect (pharmacodynamics). The models are composed of individual parameters (eg, clearance, distribution volume, effective concentration for 50% of maximal effect). Because of the complex interaction of the parameters, it is difficult to draw conclusions about drug behavior from a single parameter.

5. Because it is a mathematically based discipline, pharmacokinetics–pharmacodynamics is a distinctly unpopular subject among clinical anesthesiologists. This unpopularity is ironic considering that there is no medical specialty for whom the accurate prediction of the time course and magnitude of drug effect is more important (anesthesiologists produce profound, potentially dangerous drug effects that must be "turned on and off" in a rapid fashion).

6. Fortunately, the clinical implications of pharmacokinetic–pharmacodynamic models can be easily understood and conveyed through the use of computer simulation. Using computer simulation, a proposed dosing scheme can be "input" into a pharmacokinetic–pharmacodynamic model, producing a "picture" of the drug levels and drug effects that are expected to occur. These pictures (ie, computer simulations) are intuitively understandable and are easily applied to clinical situations.

7. The "biophase" is the theoretical site of drug action or "effect site" (eg, the brain, the neuromuscular junction, the spinal cord). It is important to consider drug concentrations in the biophase (and not just the plasma) because most drugs do not exert their effect in the blood. Pharmacokinetic–pharmacodynamic models account for this problem by linking the concentrations in the blood to theoretical concentrations in the biophase.

8. Because anesthetics are rarely administered alone (ie, anesthesia is usually at least a 2-process consisting of an analgesic and a sedative), characterizing the interaction between drugs is also an important goal of pharmacokinetic-pharmacodynamic study. Most anesthetics commonly used in combination, such as fentanyl and propofol, interact in a profoundly synergistic way (eg, where 2 + 2 = 7 or more …) so that much less of each drug is required (compared with the doses necessary when the drugs are used alone). Drug interaction models using "response surface" methods can be used to visualize these synergistic interactions and identify optimal dosing regimens.

9. Anesthesiologists have long recognized the need to adapt their anesthetic to account for differences in demographic factors and disease processes that influence drug disposition or effect. Comorbidities such as obesity, blood loss, presence of opioid tolerance, and differences in age are referred to as covariates. Covariates are descriptors of demographic factors or pathophysiologic states that impact anesthetic drug behavior.

Clinical pharmacology is the science of predicting the magnitude and time course of drug effect. Given that anesthesiologists and other anesthesia providers spend their days administering low therapeutic index agents, clinical pharmacology is perhaps more important to anesthesiology than any other specialty. From a practical aspect, the ultimate goal of clinical pharmacology is to provide anesthesia practitioners with the information they need to make rational decisions about the selection and administration of anesthetics.

Anesthesia and reanimation necessitate a standard of precision and accuracy in drug administration not required in most areas of clinical medicine. Practitioners must profoundly depress the central nervous system to maintain the anesthetized state but then rapidly reanimate patients after an operation is complete. Although overdosing every patient within the constraints of acceptable hemodynamic variables is one approach to ensuring patients are adequately anesthetized, it comes at the cost of slow emergence from anesthesia, among others. Clinicians must therefore target drug levels that are within a relatively narrow therapeutic window to achieve the competing clinical imperatives of adequate anesthesia (without toxicity) and rapid emergence.

After an agent is selected (presumably because the agent's pharmacologic profile is well suited to the proposed application), the next challenge is the formulation of a scientifically grounded dosing strategy. What does the anesthetist need to know? Table 40-1 catalogs some of the important considerations in determining the proper drug administration scheme in the context of anesthesia practice.

Anesthesiologists have long recognized that conventional, often simplified approaches to describing drug behavior such as "half-life" or "peak and trough" are not useful in answering these questions.1 Perhaps no other specialty in medicine is more dependent on accurate predictions of a drug's pharmacokinetic and pharmacodynamic profile than is anesthesiology. Most settings in clinical medicine do not require immediate onset and rapid offset of pharmacologic effect. When an internist prescribes an oral medication for treatment of hypertension, for example, the fact that several days may be required for the development of a steady-state level of drug effect is of little consequence. The anesthetist, however, must rely on drugs with rapid onset and predictable offset of effect to ensure maintenance of an anesthetic state with return of responsiveness and other vital function at the appropriate time. In sum, an understanding of fundamental pharmacokinetic and pharmacodynamic concepts is of critical importance to the clinical practice of anesthesiology.

Modern clinical pharmacology techniques and concepts provide the scientific foundation to answer the questions listed in Table 40-1. Clinical pharmacologists have created tools using principles of pharmacokinetics and pharmacodynamics to construct models that predict drug behavior. Unfortunately, the often puzzling mathematical manipulations involved in estimating pharmacokinetic and pharmacodynamic parameters and the complex mathematics required to build models that predict drug behavior have made these techniques distinctly unpopular among most practitioners. Most practitioners have been slow to integrate the intricacies of kinetics and dynamics into their clinical practice and instead rely on training and experience to determine dosing. Thus, the determination of the proper dose in clinical anesthesia is often little more than a sophisticated guess based on the anesthesiologist's gestalt impressions.

So why bother with clinical pharmacology? It is because this discipline provides the core scientific foundation for optimizing doses of intravenous (IV) anesthetics and should be a part of every practicing anesthetist's knowledge base. Fortunately, advances in pharmacologic computer simulation have revolutionized the way we apply complex pharmacokinetic and pharmacodynamic models. Simulations can be used to create meaningful pictures of drug behavior that are useful when considering the questions described in Table 40-1. Pharmacokinetic and pharmacodynamic model simulations get beyond the clinically unappealing mathematics associated with this area, providing an intuitively interpretable picture of drug behavior that clinicians can grasp and apply in day-to-day practice. Relying almost entirely on simulations (and not math), the purpose of this chapter is to empower practitioners with key concepts in pharmacokinetics and pharmacodynamics that influence the answer to: "What is an appropriate dose for my patient?"

Pharmacokinetics

To develop a pharmacokinetic model, clinical pharmacologists administer a drug and then repeatedly measure drug levels until the concentration is undetectable. The raw data are drug concentrations over time (Fig. 40-1). Using computerized pharmacokinetic tools, an equation is fit to the raw data. The equations used are simply a mathematical expression of the shape of the concentration versus time curve. The equations comprise parameters such as fractional coefficients and rate constants that are not much use to most providers.

To make parameters more meaningful, these equations are often "re-parameterized" (converted) in terms of distribution volumes and clearances or intercompartmental micro rate constants. Distribution volumes and clearances are used to create compartmental models that provide a schematic representation of drug behavior (Fig. 40-2A).

Unimportance of Individual Parameters

It is difficult for clinicians to take advantage of compartment models and the mathematical equations that represent them in a clinical setting. Consideration of individual volumes of distribution or clearance parameters in formulating a dose of anesthetic that accounts for the complex interplay of the parameters is impossible for humans to do in real time.

As an example of this complexity, consider the illustration in Fig. 40-2B. As a metaphor of a compartment model, tracking loans and bank accounts over time provides insight into what is required to estimate drug concentrations in the body over time. Consider a person with an income of $2900 that is deposited into his or her bank account each month. From this bank account, there are 2 monthly withdrawals to pay for credit card debt and mortgage payments. The credit card debt has a high interest rate (18%) with a debt load of $10 000. The mortgage has a low interest rate (5.5%) and a balance of $100 000. The monthly payments to each are $200 and $900 to the credit card and mortgage companies, respectively. In addition, each month, there is a monthly expenditure of $1000 for savings and daily living operating costs. Assuming no other expenses, how much money will be in the bank (central compartment) in 21 months? It is clear that without the use of a computer, calculating the answer to this question is impossible. Hence, using compartmental models without computer support makes them not very useful in a clinical environment.

Importance of Simulations

The real power from these pharmacokinetic parameters and compartment models comes through simulation. Because little insight into a drug's pharmacokinetic profile can be gleaned from simple inspection of its multicompartment pharmacokinetic parameters, computer simulation of the expected rise and fall of drug concentrations using a drug's pharmacokinetic parameters has assumed an important role in modern pharmacokinetic research and analysis. Making use of population pharmacokinetic parameters estimated in research studies, computers can be programmed to simulate the concentration versus time profile that results from any combination of boluses or continuous infusions. Although such simulations are subject to certain limitations, they are intuitive graphic representations of the time course of drug concentration.2

For example, Fig. 40-3 illustrates a set of simulations of propofol using pharmacokinetic parameters from the literature.3 In these simulations, the temporal profiles of plasma propofol levels that result from various bolus doses are plotted over time. From these plots, it is possible to visualize the peak plasma propofol concentration attained and the time propofol levels remain present in the plasma.

Without the aid of a computer, these simulations would be impossible. Although the simulations are limited by the quality of the original research from which the pharmacokinetic parameters were estimated and the inherent variability of pharmacokinetic parameters from patient to patient, the simulations are nonetheless graphic representations of a drug's expected clinical pharmacokinetic profile and provide an excellent framework within which to formulate a rational dosing strategy.

Context-Sensitive Half-Times

Computer simulation has also been useful in demonstrating how estimating drug behavior based on individual kinetic parameters such as terminal half-lives can be misleading.1 Clinicians have traditionally relied on terminal half-lives as a reflection of the duration of drug action when in fact terminal half-life alone is not a very useful pharmacokinetic parameter.4 To that end, techniques have been used to predict the time necessary to achieve a 50% decrease in drug concentration after termination of a variable lengths continuous infusions. Simulation approaches have been developed to provide a context-sensitive half-time or 50% decrement time in which context refers to the duration of a continuous infusion.5 Such simulations are intended to provide more clinically relevant parameters than "half-life" or "volume of distribution."6

Fig. 40-4 is a graphical representation of context-sensitive half-times for selected opioids and sedatives using parameters from the literature.3,7-11 As can be appreciated in part A, clinically relevant information is easily gleaned from the plot that compares the context sensitive half-times for propofol, etomidate, and midazolam. Here propofol and etomidate have a more rapid decrement time than midazolam for any continuous infusion of duration longer than 30 minutes. This illustrates why a prolonged continuous infusion of midazolam is a poor choice if rapid emergence is an important anesthetic goal. By contrast, propofol and etomidate have much more forgiving profiles for prolonged continuous infusions. Even for infusions that last more than 8 hours, for example, the time required for propofol to decrease by 50% is less than 20 minutes. This feature of propofol makes it an attractive drug when delivering prolonged IV anesthetics. Etomidate, although it has a favorable pharmacokinetic profile for prolonged infusions compared with midazolam, has other features that make it unattractive for prolonged use such as nausea and vomiting, hemolysis, and adrenal suppression.12

With regard to opioids, sufentanil appears to have more favorable pharmacokinetics for infusions lasting less than 8 hours compared with fentanyl when the goal is to achieve a rapid 50% decrease in concentration. This difference can be explained by the fact that sufentanil's pharmacokinetic model has a large, slowly equilibrating peripheral compartment that continues to fill after termination of an infusion, thus contributing to the faster decrease in sufentanil central compartment concentration. In other words, central compartment sufentanil concentrations (which is the compartment that drives drug effect) decrease rapidly after an infusion of less than 8 hours is stopped because of continued elimination and distribution.5,13

Unlike sufentanil, fentanyl exhibits an early time-dependent increase in the context-sensitive half-time. Although fentanyl would be a poor choice for clinical situations in which a rapid decrease in concentration after infusion termination is desirable, in clinical scenarios in which prolonged opioid effect is the goal, fentanyl might well be the drug of choice. For example, fentanyl is well suited for cases after which the patient's trachea will remain intubated for a period of time after the procedure to promote a gradual emergence from anesthesia and a long-lasting level of significant analgesia.

Note that for cases of very brief duration, the context-sensitive half-times for sufentanil and fentanyl are nearly identical. Thus, for brief applications, when the opioid is administered by infusion (or by frequent small bolus doses), there would not be any substantial differences among these drugs in the time to a 50% decrease in concentration after stopping a continuous infusion.

Remifentanil, in contrast to both sufentanil and fentanyl, has a very short context-sensitive half-time for infusions of any duration. This feature may be attractive during procedures associated with varied surgical stimuli during which giving longer-acting opioids such as fentanyl or sufentanil to blunt the response to noxious stimulation may prolong emergence or increase the chance of unwanted respiratory depression. Remifentanil's short context-sensitive half-time may be useful in cases in which timely postoperative neurologic assessments are warranted to rule out developing neurologic deficits after neurosurgical procedures. Remifentanil, on the other hand, may be a poor choice as a prolonged infusion in cases associated with significant postoperative pain (ie, total hip arthroplasty). In this setting, when a continuous infusion of remifentanil is terminated, without the addition of a "transition" analgesic, patients can become remarkably uncomfortable when remifentanil plasma levels wane.

Although the 50% decrease in time is an improvement compared with the use of the terminal half-life for estimating how anesthetic drugs will behave, it is not always relevant. Depending on the dose and pharmacokinetic features of a particular drug, the 50% decrement time may not adequately describe drug behavior that is of clinical interest (eg, when will the analgesic effect of my IV opioid infusion dissipate after the infusion is terminated?). To get beyond this limitation, other decrement times can be used, such as the 20% or the 80% decrement time, to tailor the pharmacokinetic description of drug behavior to a clinical endpoint of interest.6,13

Importance of Biophase

Biophase refers to the equilibration delay between peak drug levels in the blood or plasma and peak drug effect. The time lag (or hysteresis) between peak concentration in the plasma and peak drug effect is a function of drug movement into and action within the effect site.14-16 The effect site represents a theoretical space without any definitive anatomic analog where drug exerts its effect. The lag time represents a summation of events that can impact the onset of pharmacologic effect such as drug diffusion to the effect site, receptor binding, and so on (Fig. 40-5A). Hysteresis is important to account for when forecasting drug effect. It is particularly important to consider when simulating bolus injections of drug; for long infusions, the time lag assumes less importance because the effect site and plasma are generally closer to equilibrium.

Consider again a 2-mg/kg bolus of propofol and the observed changes in a processed electroencephalogram (EEG) parameter such as the bispectral index scale (BIS). The maximal decrease in the BIS lags behind the peak plasma propofol level (Fig. 40-5B). The time delay, or hysteresis, represents the time required for propofol to go from the plasma to the site where it exerts an effect. In other words, most drugs do not work in the plasma.

For many drugs, including propofol, the equilibration delay between peak concentration in the plasma and peak effect has been characterized by estimating a parameter called the keo. The keo represents the rate constant for elimination of drug from a virtual compartment called the effect site.14,17 (Fig. 40-5C). The effect site concept is schematically represented as an additional compartment in the 3-compartment model (Fig. 40-6). When the keo parameter is available for a drug, theoretical effect compartment concentrations can be simulated along with plasma concentrations, thus making the effect of the time lag easily appreciated. In essence, the time course of effect site concentrations should accurately predict the time course of drug effect.

One of the useful clinical features of simulating the effect site concentration is that the time required to reach peak effect can be easily visualized. Fig. 40-7 presents simulations of the effect concentration over time for selected sedatives and opioids after commonly used doses for each drug. It is readily apparent that some IV anesthetics reach their peak effect site concentrations for a given dose much later than others (Table 40-2).

Midazolam, for example, requires up to 9 minutes to reach peak effect in contrast to propofol and etomidate. This may explain why the early use of midazolam for endoscopic procedures was associated with adverse events related to excessive sedation.18-20 One misleading characteristic of midazolam is that it has a rapid onset of effect (ie, some effect is apparent soon after injection), but the latency to peak effect is slower relative to some other sedatives. When a desired effect has not been reached within 90 seconds, practitioners may be tempted to administer additional midazolam. This may lead to a peak effect site concentration that is excessive and delayed in presentation compared with other sedative hypnotics. This underscores the importance of using caution when administering large doses of IV midazolam.

By contrast to midazolam, etomidate and propofol, given as a bolus dose under normal hemodynamic conditions, reach their peak effect within 2 minutes after administration and then quickly dissipate. With these agents, if a desired effect has not been achieved within 120 seconds, practitioners may be inclined to give additional drug knowing that the previous doses are already waning.

With regard to opioids, after a bolus dose, the time required to reach peak effect site concentrations for remifentanil is considerably quicker than that of fentanyl and sufentanil. The clinical implications of this difference may be considerable when administered to patients breathing spontaneously. Remifentanil by bolus injection may lead to a more pronounced respiratory depression compared with equipotent doses of sufentanil or fentanyl. With the slower onset of effect for fentanyl and sufentanil, blood PCO2 levels are given time to rise, providing some offset to the respiratory depression associated with opioids. With remifentanil, the onset of effect outpaces the accumulation of blood PCO2, leading to more pronounced respiratory depression.21-23

Pharmacodynamics

Pharmacodynamic models have been constructed to describe the relationship between drug effect site levels and drug effect. Some important features of pharmacodynamic models are presented in Fig. 40-8. In this figure, a schematic illustrates how drug, when it reaches the effect site, interacts with a receptor to produce effect (Fig. 40-8A). This process is typically characterized graphically using a sigmoidal curve (Fig. 40-8B). Parameters used to describe the pharmacodynamic model (ie, the sigmoid curve) include the C50 and γ. The C50 represents the effect site concentration at which 50% of the maximal drug effect will be elicited, and γ represents the slope of the concentration–effect curve. The most important part of the sigmoid curve, the steep section of the curve, represents the dynamic range of drug effect. The dynamic range charts the concentration–effect relationship from E0 (baseline effect) to Emax (maximal effect). In this region, small changes in drug concentration lead to large changes in drug effect. This is the region of particular interest to anesthesia providers. Increasing effect site concentrations beyond the dynamic range leads to minimal changes in drug effect. The mathematical expression that uses the C50 and γ to estimate drug effect is presented in Fig. 40-8C.

A surfing analogy is helpful in conceptualizing the application of pharmacodynamic models to rational drug administration.24 Just as a surfer attempts to ride just in front of the crest of a wave to slide down the front edge of the wave, practitioners attempt to maintain their anesthetic drugs at effect site concentrations near the "crest" of the concentration–effect relationship (Fig. 40-8B). In so doing, clinicians maintain significant drug effect but can "slide down" the concentration curve to promote recovery at the end of an anesthetic. From an efficacy and toxicity perspective, there is no advantage of being on the flat part of the concentration–effect relationship near Emax.

As an example, consider 3 bolus doses of propofol (0.5, 2, and 7 mg/kg). The resultant effect site concentrations for each bolus are presented in Fig. 40-9. Using a pharmacodynamic model for propofol, the effect of these boluses is plotted on a sigmoid curve. The C50 for this curve (1.8 μg/mL) represents the effect site concentration at which there is a 50% probability of loss of responsiveness.25 Dosing regimens that maintain drug concentrations along the lower left portion of the sigmoid curve (ie, below the wave) are too low to be effective, as illustrated by the low-dose propofol bolus (0.5 mg/kg). Dosing regimens that maintain drug concentrations to the right side of the sigmoid curve (ie, before the wave breaks) are excessive and may produce unwanted hemodynamic depression or prolonged recovery. In this region, increasing drug concentration does not increase drug effect. This phenomenon is illustrated by the 7-mg/kg bolus dose of propofol. The ideal dosing strategy targets the upper portion of the steep part of the concentration–effect relationship: A concentration that produces considerable drug effect but from which drug effect will recover quickly when drug administration is terminated as illustrated by an intermediate dose of propofol (2 mg/kg).

One practical aspect of anesthesia care that makes pharmacologic "surfing" difficult to do is that the concentration–effect relationship is not consistent across various stimuli. For some stimuli, much less drug is required to achieve a desired effect compared with other stimuli. For example, skin incision is typically less noxious than laryngoscopy.26-28 In an effort to characterize how different stimuli vary, numerous pharmacodynamic models have been built for selected stimuli.

A schematic, presented in Fig. 40-10, plots out the relative difference between pharmacodynamic models for opioids across various stimuli. Effect measures used to characterize opioid behavior include mild to moderate analgesia (ie, blunting response to a noxious stimulus), respiratory depression, significant analgesia (ie, blunting response to laryngoscopy), and suppression of EEG activity. As expected, the pharmacodynamic relationship for each stimulus is very similar (a sigmoid curve), but the C50s are shifted from left to right with increasing stimulus. Several important points flow from this schematic, especially for the meticulous clinician who seeks to "surf" near the ideal Ce and thus achieve the desired drug effect without administering a relative overdose. These include (1) significant analgesia is required to blunt the response from laryngoscopy compared with other stimuli encountered in the operating room; (2) the C50 for the effect site concentration necessary to blunt a response to laryngoscopy is higher than the C50 for respiratory depression; (3) dosing anesthetics requires an appreciation of ongoing and anticipated stimuli; and (4) there is a window in opioid levels between which one can achieve analgesia yet avoid excessive respiratory depression. Investigators have demonstrated that the C50 for analgesia is approximately 30% of the C50 for respiratory depression.26 This may be especially important during emergence from anesthesia.

Importance of Combining Kinetic and Dynamic Models

To depict a patient's response to a dose of drug, it is necessary to combine pharmacokinetic–pharmacodynamic models and provide a quantitative description of each. Because most drugs do not act in the blood, pharmacokinetic and pharmacodynamic models must be linked so that concentrations in the plasma can be translated into effect site concentrations and thus drug effect. One approach to visualize the link between the pharmacokinetics and pharmacodynamic models is to plot a horizontal line on the pharmacokinetic plot of effect site concentrations over time that represents the C50 for a desired drug effect. Two key points of interest are now easily visualized through computer simulation: the time to onset of effect and the duration of drug effect.4,29,30

For example, using linked pharmacokinetic and pharmacodynamic models, the simulations of propofol bolus doses ranging from 1 to 2.5 mg/kg presented in Fig. 40-3 can be re-performed for effect site concentrations rather than plasma concentrations (Fig. 40-11). To visualize the onset and duration of effect, the C50 for loss of responsiveness is overlaid on the simulations of propofol effect site concentrations. Several investigators have estimated propofol effect site concentrations required for loss of responsiveness with C50s ranging from 1.8 to 2.4 μg/mL.25,31-33 For simulation and discussion purposes, the C50 for loss of responsiveness to be used in these simulations and throughout the remainder of this chapter will be 1.8 μg/mL. A summary of the time to onset and duration of effect for each propofol bolus dose is presented in Table 40-3. It is important to point out that this C50 for loss of responsiveness, similar to all pharmacokinetic and pharmacodynamic parameters, is a population estimate for the typical patient and as such is subject to intersubject variability and will not perfectly predict unresponsiveness in individuals.

Guided by simulations, the clinical implications of a given propofol dosing strategy become apparent. For example, using a higher dose of propofol as part of a rapid sequence induction, if hemodynamically tolerable, may prove useful in minimizing the time when the airway is unsecured because of faster onset of effect. By contrast, a lower dose may be useful when providing anesthesia for brief procedures associated with a noxious stimulus, such as a retrobulbar block, during which only brief unresponsiveness is required. When using the lower dose, however, practitioners need to recognize that it will take up to 50 seconds longer to achieve loss of responsiveness.

Simulations may also be useful in illustrating how drugs behave in relation to an unwanted toxic effect. Consider a plot of fentanyl effect site concentrations for various doses of fentanyl and the concentration at which probability of substantial respiratory depression exceeds 50% (ie, somewhere around 3.5 ng/mL) in 60-, 80-, and 100-kg people.34 (Fig. 40-12). In this simulation, the bolus of 100 and 150 μg of fentanyl avoids respiratory depression for each weight. A 250-μg bolus, however, exceeds the effect site concentrations associated with a 50% probability of respiratory depression for people weighing 60 and 80 kg for 7 and 13 minutes, respectively. According to these simulations, persons weighing 100 kg or more briefly approach the fentanyl effect site concentration associated with a 50% probability of respiratory depression but do not exceed it.

Linked pharmacokinetic–pharmacodynamic models can also provide insight into how drugs behave after termination of continuous infusions. Factors that influence the duration of effect after termination of an infusion include the infusion rate and the duration of the infusion. For example, consider the time required to regain responsiveness after termination of a continuous infusion of propofol. Fig. 40-13 shows a series of simulations of the resultant propofol effect site concentration after continuous infusions at rates between 50 and 150 μg/kg/min for 1, 2, and 3 hours. As with the previous simulations of bolus dosing of propofol, the plot also charts the propofol effect site concentration associated with a 50% probability of loss of responsiveness (1.8 μg/mL).25

The area of interest is the segment of time from when the infusion is terminated until the propofol effect site concentration falls below the level associated with a loss of responsiveness. The results of these simulations are summarized in Table 40-4. When the infusion was terminated, longer propofol infusions at equivalent dosages required more time for propofol levels to drop below a level associated with unresponsiveness. As expected, once the infusion was terminated, higher propofol infusion rates required more time to drop below a level associated with unresponsiveness. With higher infusion rates, the time to regaining responsiveness can be quite long (ie, up to 13 minutes).

Conceptualizing this clinical issue in terms of compartmental models, these simulations illustrate how during prolonged infusions propofol gradually "fills" the slowly equilibrating peripheral compartments (extravascular tissues). When the infusion is terminated, the propofol in the peripheral compartments serves as a reservoir to maintain plasma propofol concentrations, thereby slowing the decline in drug effect. This may be clinically evident as a delay in emergence. These simulations also suggest that running a propofol infusion at a single continuous rate throughout an anesthetic will lead to propofol effect site levels that may be higher than desired. A prudent approach to dosing prolonged propofol infusion therefore is to reduce the infusion rate over time to minimize drug accumulation. This accumulation phenomenon emphasizes the advantage of target-controlled infusions (TCIs) that instead of delivering drug at a set rate (ie, μg/kg/min) deliver drug to achieve and maintain a set effect site concentration (ie, 2.5 μg/mL for propofol). In so doing, they avoid the accumulation of drug yet maintain an appropriate drug level for the desired duration. These types of infusion systems are widely used throughout the world, yet unfortunately have not attained regulatory approval within the United States.24

This form of analysis may be also useful to use when exploring the duration of analgesic effect from fentanyl after surgical procedures of various durations. A common practice is to administer intermittent boluses of fentanyl starting at induction and then throughout a general anesthetic with the goal of providing perioperative analgesia during and after a stimulating procedure. Consider an anesthetic where fentanyl, in addition to a potent inhaled agent, is given as a bolus (3 μg/kg or ˜200 μg for a 70-kg patient) on induction and then intermittently (ie, 1.5 μg/kg or ˜100 μg for a 70-kg patient every 20 minutes) throughout a surgical procedure associated with significant painful stimuli.

Simulations of this fentanyl dosing regimen for anesthetics lasting 1, 2, and 3 hours reveal several clinical points of interest (Fig. 40-14). For a 1-hour anesthetic with an induction dose of 200 μg of fentanyl followed by 2 supplemental 100-μg doses during the anesthetic, the total fentanyl dose administered is 400 μg (8 mL). With this dosing regimen, the fentanyl effect site concentration intermittently rises above the fentanyl effect site concentration associated with analgesia (1.6 ng/mL)35 but rapidly drops below the analgesic level after the anesthetic is terminated.

Simulations of the 2- and 3-hour anesthetics using a similar dosing scheme result in total fentanyl doses of 700 and 1000 μg (14 and 20 mL). With the longer duration, the fentanyl concentration rises above the effect site level associated with analgesia. In this simulation, what becomes apparent is that repetitive doses increase the fentanyl concentration at the end of a 3-hour anesthetic; the resulting effect site concentration of fentanyl is much higher than after the 1-hour anesthetic. After termination of the anesthetic, the time required for the fentanyl effect site concentrations to decrease below the level associated with analgesia was 9 and 57 minutes for a 2- and 3-hour anesthetic, respectively. Similar to what was learned in the simulations of the context-sensitive half-time for fentanyl, prolonged infusions or repetitive dosing of fentanyl over prolonged periods can lead to high fentanyl effect site concentrations. High fentanyl levels may be of clinical benefit (ie, adequate analgesia after a pain surgical procedure) or detriment (prolonged respiratory depression) in the early postoperative period. These simulations demonstrate the potential benefit of visualizing fentanyl effect site concentrations in real time when caring for patients in whom the anesthetic goals upon termination of a general anesthetic include adequate analgesia yet avoidance of unwanted respiratory depression. At present, many investigators are developing tools to allow clinicians to visualize both drug concentrations and effects over time.

One very important limitation of these simulations is that painful surgical procedures are rare if ever performed with fentanyl alone. Other anesthetics, such as potent inhaled agents or IV sedatives, are used as well. Inhaled agents and IV sedatives accentuate the analgesic effect of opioids in a synergistic fashion. This limitation is addressed in the following section. Thus, these simulations chart the effect of fentanyl alone and at best are representative of analgesic effect after other anesthetics that have been coadministered with the fentanyl have been allowed to dissipate. Hence, the segment of this simulation set reflective of the actual analgesic state are the fentanyl effect site concentrations well beyond (ie, 20-30 minutes) the termination of the anesthetic.

Drug Synergism

Anesthetics are typically administered as a combination of several different types of drugs to achieve a desired complete anesthetic. Anesthesiologists have long appreciated that the administration of 1 type of drug may enhance and prolong the effect of another type of drug. The concept of minimum alveolar concentration (MAC) reduction is well established and used to describe how the addition of an IV opioid reduces the amount of a potent inhaled agent required to achieve and maintain a desired MAC. This line of thinking is especially important when trying to avoid unwanted side effects associated with high doses of a single anesthetic such as hemodynamic depression, prolonged emergence from anesthesia, or persistent respiratory depression.

As clinical pharmacologists have become more sophisticated in their approach to characterizing the interaction between different anesthetic drug types, several tools have been developed to describe drug–drug interactions. One simple tool characterizes drug–drug interactions as additive (ie, 2 + 2 = 4), antagonistic (eg, 2 + 2 = 1), or synergistic (eg, 2 + 2 = 8). These can be graphically illustrated to allow visualization of the extent of antagonism or synergism present between 2 drugs (Fig. 40-15). These drug interaction plots present the effect site concentrations for 2 drugs that are necessary to achieve specified drug effects. In this schematic a desired effect is achieved at point A on the horizontal axis and point B on the vertical axis. The points labeled a and b represent the concentrations of both drugs required to achieve a similar effect to either drug A or B alone. As can be appreciated, there are an infinite number of drug–drug combinations that result in a similar drug effect. The line that runs through all the possible drug–drug concentration pairs and connects points A and B is known as an isobologram. The isobologram represents drug concentration pairs that would result in the same drug effect when the 2 drugs are used alone or in combination.

With drug–drug interactions that are additive, the isobologram is a straight line indicating that as drug A increases, proportionally less of drug B is required to achieve the same degree of effect (Fig. 40-15A). With drug–drug interactions that are synergistic, the isobologram is a curved line bowing toward the origin of the graph (Fig. 40-15B), indicating that when both drugs A and B are used, much less of both is required to achieve the same degree of desired effect. Conversely, with drug–drug interactions that are antagonistic, the isobologram is a curved line bowing away from the origin of the graph (Fig. 40-15C), indicating that when both drugs A and B are used, much more of each is required to achieve the same degree of desired effect. Most interactions between anesthetic drugs are synergistic.

Common anesthetic goals can be characterized in terms of isobolograms. For example, isobolograms can be used to plot effect site concentration pairs necessary to achieve a 50% or 95% probability of achieving a desired drug effect (ie, no response to laryngoscopy) or when a worrisome side effect may occur (ie, onset of respiratory depression). Using these isobolograms, it is possible to explore through computer simulations how well an anesthetic performs in terms of meeting desired anesthetic goals.

For example, consider a schematic of a combined anesthetic technique using drugs A (an opioid) and B (a sedative) that are known to have a synergistic interaction (Fig. 40-16). Drug delivery that yields concentration pairs to the left of 50% isobole for no response to painful stimuli do not meet analgesic goals. With drug pairs in this region, patients will most likely respond to painful stimuli. Drug delivery that yields concentration pairs between the 50% and 95% isoboles will likely meet analgesic goals, but some patients may respond to a painful stimulus. In this region, after the anesthetic is terminated, the analgesic effect will quickly dissipate. With drug delivery that yields concentration pairs to the right of the 95% isobole, a patient is very likely to be unresponsive to painful stimuli, and increasing the anesthetic target will not substantially increase the probability of adequate effect. A key point with regard to this region of the isobole plot is that administering additional anesthetic will not provide more analgesia but rather only prolong the time required for drug effect to dissipate, especially if the resultant concentrations have far exceeded the 95% isobole. A prolonged duration of effect may be a desired outcome (ie, adequate analgesia after a surgery associated with significant postoperative pain) or an undesired outcome (ie, delayed emergence after a prolonged surgery).

Because visualization of these complex interactions is currently not feasible in a clinical setting and clinicians are justifiably more frequently concerned with preventing awareness while patients are under anesthesia or exaggerated responses to painful stimuli than promoting rapid emergence from anesthesia, most anesthetic techniques target concentration pairs that are substantially above the 95% isobole. In sum, to avoid unwanted consequences of light anesthesia, most practitioners tend to administer relatively excessive doses of anesthetics. For the majority of anesthetics, this practice is well tolerated. However, in patients sensitive to adverse consequences of excessive anesthesia, administering an anesthetic beyond the 95% isobole may be associated with adverse consequences.

In order to visualize the concentration interaction 2 drugs have on a particular drug effect such as analgesia, recent work has focused on the development of 3-dimensional plots called response surfaces. Response surface portrays the concentration–effect relationship for each drug individually (similar to a simple pharmacodynamic model) as well as paired concentration–effect (drug–drug interaction). Plotted on these surfaces are the effect site concentrations for 2 drugs (ie, an opioid and a potent inhaled agent) over a range of concentrations of clinical interest and a measure of drug effect. The measure of drug effect is typically presented as a probability ranging from 0 to 1 (ie, the probability of loss of responsiveness with a selected drug pair is 0.8). Recently, many investigators have characterized response surfaces for a variety of drug–drug pairs and clinical relevant drug effects in human volunteers and patients. For example, surfaces have been developed for analgesia, loss of responsiveness to verbal and tactile stimulation, and laryngoscopy for propofol and remifentanil25,33,36-41 and sevoflurane and remifentanil.42

To illustrate the visual power of response surfaces, consider a simulation of 90-minute total IV anesthetic using continuous infusions of propofol and remifentanil (Fig. 40-17). Simulated infusion rates are consistent with dosing recommendations for each drug. The resultant effect site concentrations for both drugs are plotted over time. With a bolus dose before starting a continuous infusion, therapeutic levels for each drug are quickly achieved and then maintained throughout the anesthetic. To visualize the combined analgesic effect of remifentanil and propofol, this anesthetic technique has been plotted out on a previously established response surface for analgesia.25 (Fig. 40-18).

With the concentration pairs of remifentanil and propofol plotted for the entire anesthetic on this response surface, 2 important features of the anesthetic technique are easily observed. First, remifentanil and propofol have a pronounced synergistic interaction. When given together, much less of both drugs is required to achieve a desired analgesic effect than if given individually. This is best illustrated by the 50% and 95% isoboles in Fig. 40-18B. Here, the isoboles bow inward and reveal that propofol, when dosed to produce an effect site concentration of 1 μg/mL, decreases the remifentanil needed to meet an analgesic goal by more than half. Of note, a propofol effect site concentration of 1 μg/mL by itself is typically enough to sedate someone but not enough to render them unconscious.

Second, the dosing used in this total IV technique results in a combined drug–drug analgesic effect that for a majority of the anesthetic exceeds the 95% isobole for analgesia. The trajectory of the dosing regimen is color-coded into 3 segments: the induction (ie, bolus doses of both drugs), maintenance (continuous infusions of both drugs), and emergence (termination of both drug infusions). It is interesting to observe that the maintenance phase of the anesthetic exclusively occupies the relatively flat region on the surface. Maintaining an anesthetic that is so far removed from the 95% isobole is essentially unnecessary. With regard to analgesia, in this region, additional anesthetic would not be expected to increase the analgesic effect. In the case of remifentanil and propofol, this practice may not be overly worrisome because both drugs exhibit a rapid decline in drug effect when infusions are terminated. With other opioid–sedative drug pairs, however, overdosing an anesthetic onto the flat portion of the response surface will slow the decline of drug effect in most instances.

One of the challenges of drug synergism is to quantify the magnitude of drug–drug interactions in a clinically meaningful manner that will allow clinicians to reliably modify their dosing regimens to optimize a desired drug effect. Although response surfaces and isoboles provide useful tools for characterizing drug interactions, their clinical utility when delivering an anesthetic in real time has yet to be explored. Suspecting that anesthesia providers may have difficulty interpreting response surfaces and isoboles, investigators have explored other means of visually presenting synergistic effects in real time. Recent work has focused on plotting the contribution of simultaneously administered sedatives or potent inhaled agents and opioids to the overall analgesic and sedative effects over time.43 With this approach, the overall analgesic effect, for example, is plotted over time as a function of the C50 for no response to a painful stimulus. Using this technique, the total IV anesthetic introduced in Fig. 40-17 has been replotted in terms of the contribution of remifentanil alone (blue) and remifentanil and propofol (green) to the overall analgesic effect (Fig. 40-19).

To gain a perspective of how synergistic these 2 drugs are with regard to analgesia, consider the contribution of remifentanil without the propofol to analgesia. The bolus and infusion of remifentanil rendered an effect site concentration that approached half of the effect site concentration (4-5 ng/mL) necessary to reach the C50 for analgesia (8.8 ng/mL).25 By contrast, when added to the bolus and continuous infusion of propofol, the cumulative effect of both drugs exceeds the C50 for analgesia 8-fold during most of the anesthetic.

An important limitation of these simulations used to describe drug synergy is that they are inherently limited by pharmacokinetic and pharmacodynamic parameter variability within the population that was used to estimate them.44,45 Population pharmacokinetic parameters should thus be viewed as an estimate of the population's typical parameters, recognizing that each individual will likely vary from the population parameters to some degree.46 Some patients, of course, will vary greatly from the population mean. For these patients, the simulations are admittedly of little value. For most patients, however, the prediction based on the population pharmacokinetic and pharmacodynamic parameters are a good starting point for initial therapy. Adjustments in dosing scheme are then made on the basis of patient response.

Visualizing Anesthetic Drug Effects in Real Time

Anesthesiologists often consult scientific journal articles and textbooks that describe drug behavior to develop general guidelines when formulating rational dosage schemes (ie, older patients require less propofol). However, the available kinetic-dynamic information is much more comprehensive and is likely to be of more value when applied in real time at the point of care. Most information regarding anesthetic drug behavior exists in the form of pharmacokinetic, pharmacodynamic, and response surface interaction models. These models are often complex and mathematically oriented and unfortunately appear in scientific journals that are not intended for practicing anesthesiologists. This means that perhaps only a small part of the information describing anesthetic drug behavior revealed in clinical pharmacology studies is ever translated into clinical practice. Until recently, the mathematical complexity of these models has precluded their practical introduction into the operating room.

Research is now being conducted to bring anesthetic pharmacologic models to the operating room through the use of drug interaction and conventional kinetic-dynamic models that visually display the time course of not only the effect site concentrations but also how anesthetic drugs interact and the resultant drug effects over time.41,43,47-49 This technology automatically acquires from infusion pumps and anesthesia machines the drug doses administered by the anesthesiologist and shows the drug dosing history (bolus doses, infusion rates, and expired concentrations); the predicted drug concentrations at the site of action (past, present, and future); and the predicted drug effects, including sedation, analgesia, and neuromuscular blockade.41,42,47,50

Compared with TCI systems, visual display systems potentially represent a significant advance because they include not only pharmacokinetic predictions of drug concentrations but also pharmacodynamic predictions of the likelihood of certain anesthetic effects. Response surface models are at the core of these display systems; that is, the information these displays present is based on response surface drug interaction models.

A basic assumption of these displays is that anesthesia care providers cannot solve complex polyexponential equations in their head in real time to choose the best dose for each patient. A major goal of the drug display systems is to bring the sophisticated body of clinical pharmacology information to the point of care in a useable form in real time.51-53 Existing prototypes of these display systems provide 3 forms of information: (1) drug concentrations (plasma, end tidal, or effect site) over time, (2) drug effects (as individual drugs and the resultant effect from drugs in combination) over time, and (3) 3-dimensional or topographic views of response surfaces for various drug effects.

A major advantage of these display views is that anesthesia care providers can visually appreciate the extent to which their anesthetic technique leads to synergistic effects between drug types. For example, sedatives or inhalation agents, when coadministered with an opioid, have a large synergistic effect on analgesia and, by comparison, a much smaller synergistic effect on sedation. Another advantage of the display is that it provides a more sophisticated way to assess the effectiveness of an anesthetic beyond what is offered on many physiologic displays through the use of the MAC concept. Anesthesiologists rarely dose their anesthetic so that 50% of their patients will move during incision. They always give more! The drug display provides not only a 50% probability of achieving an anesthetic goal but a 95% probability as well. The 95% probability is much more in keeping with how anesthesiologists dose their anesthetics.

In addition, the drug display separates the components of anesthesia into sedation, analgesia, and muscle relaxation and provides a probability of response for each component. This offers anesthesiologists a practical approach to designing and modifying anesthetics to meet the dynamic changes associated with various patients and surgical procedures. Additionally, these drug displays have the potential to simulate various therapeutic decisions immediately before they are implemented, thus allowing anesthesiologists to explore the consequences of a proposed change in therapy prospectively.

Fig. 40-20 displays 2 prototype systems currently in development. Although somewhat different in terms of display format, the displays both use tabular and graphical presentation of predicted drug concentrations and drug effects, including a prediction of the synergistic interaction between hypnotics and opioids. Both systems include prediction modules that allow anesthesia care providers to explore various dosage regimens before administration to best meet the anesthetic needs of a surgical procedure. For example, it is possible to simulate the decay of drug concentrations and the projected time to recovery if the drug administration were stopped 5 minutes in the future.

Numerous challenges need to be overcome before these display systems can be adopted into widespread clinical practice. How these drug displays will improve clinical outcomes (eg, faster recovery, improved analgesia on emergence) and how they will gain clinician acceptance (eg, decreased physician workload) must be demonstrated in clinical testing before widespread adoption. Preliminary evidence suggests that they will perform reasonably well,47-49 but much work remains to be done. One of the barriers to implementation involves the anesthesia care providers' level of understanding of the scientific basis of these complex models and what can be expected from their use during clinical care. Anesthesiologists will likely require education and training to fully use the information this technology offers.

Although it is too early to predict how drug display technology will be used in future anesthesia practice, the concept provides a promising potential to bring more sophisticated clinical pharmacology knowledge to the point of care. Similar to the pharmacokinetic component of these displays already widely implemented in the form of TCI systems, it is conceivable that in the future a real-time display of the predicted pharmacokinetics and pharmacodynamics of anesthetic drugs might be found alongside the traditional physiologic vital sign monitors.54

In summary, most practitioners tend to relatively overdose patients with their anesthetic technique because the current surrogate measures of anesthetic depth (ie, changes in heart rate, blood pressure, patient movement, processed EEGs) and display tools to visualize the time course of anesthetic drug concentrations over time are inadequate. Given that variations in surgical stimulus are often difficult to predict and that pharmacologic interpatient variability is substantial, clinical experience mandates that clinicians continue to administer generous doses of anesthetics. As pharmacokinetic and pharmacodynamic display systems become more effective at reliably communicating useful information in a timely manner regarding clinical endpoints of interest, perhaps anesthesiologists will be more inclined to tailor their technique to the precise needs of each patient and the demands of a given surgical procedure or, in other words, surf the 95% isobole.

Anesthesiologists have long recognized the need to adapt their anesthetic to account for differences in demographic factors and disease processes that influence drug disposition or effect. Increasingly, the scientific rationale to support these dosing changes is taking shape. Comorbidities in these special populations such as obesity, renal failure, heart failure, liver failure, blood loss, presence of opioid tolerance, and differences in age are referred to as covariates. Covariates are descriptors of demographic factors or pathophysiologic states used to estimate the impact these states have on anesthetic drug behavior. Although a significant amount of research has been dedicated to describing covariates and how they can be used to optimize dosing, there are still many gaps in our knowledge base, making the development of guidelines and dosing recommendations for commonly used anesthetics difficult. Nevertheless, as newer IV anesthetics have expanded the array of drugs available to practitioners, covariate analysis has become an increasingly useful tool to describe potential dosing pitfalls associated with these anesthetics.

The next segment of this chapter reviews comorbidities and demographic factors that have been described with covariate analysis for selected IV anesthetics (propofol, fentanyl, and remifentanil) as an example of how to account for disease states such as obesity and blood loss and commonly occurring patient conditions such as advanced age and opioid tolerance when formulating a dosing regimen. Unfortunately, our pharmacologic database is incomplete and does not allow us to generalize these examples to all IV anesthetics. Furthermore, some of the studies have been performed in animal models and are therefore difficult to translate to clinical practice. Nevertheless, this body of work does provide some insight into how IV anesthetics behave in the presence of commonly occurring covariates and is important to consider when formulating a dosing regimen.

The clinical relevance of accounting for differences in body mass is a result of the prevalence of obesity in the Western culture. Obesity is a major public health problem throughout the developed world. Since the early 1970s, the proportion of the American population that is overweight has steadily increased.55 Among US adults ages 20 to 74 years, approximately 25% are overweight with a slightly higher prevalence among women.56 Almost 5% of US adults are morbidly obese (ie, they weigh twice their ideal body weight [IBW]). Anesthesia providers thus frequently encounter obese patients in everyday practice.

Many investigators have explored the impact of weight on drug dosing for opioids46,57-59 and sedatives60 and have offered strategies on how doses should be formulated for patients who are overweight. Nonetheless, despite the high prevalence of obesity, practitioners often formulate dosage regimens for many drugs based on total body weight (TBW). Dosing according to TBW in obese patients, however, can lead to large doses and prolonged or toxic effect. Clinicians should recognize that most studies designed to determine appropriate doses of IV anesthetics have been conducted in healthy patients or volunteers at or near their IBW. A unique attribute of IBW is that it represents an estimate of appropriate body weight based on height only. For example, a popular formula to estimate IBW is 49.9 + 0.89 × (Height – 152.4) kg for men and 45.4 + 0.89 × (Height – 152.4) kg for women.61

To illustrate the resultant weights using these formulas, IBWs for several common heights are presented in Table 40-5. It is important to point out that when using these formulas for any given height, the IBW is the same regardless of weight. Formulating a dose according to the IBW for patients of equivalent height that are 70, 100, and 150 kg would be the same. By contrast to dosing patients according to their TBW, dosing according to the IBW has the potential for significant underdosing.

To get around the limitations of IBW for dosing anesthetics, investigators have used lean body mass (LBM) instead. Advantages of LBM are that it excludes weight associated with adipose tissue and accounts for patient height and TBW. For example, LBM has been used to scale pharmacokinetic parameters (ie, volumes and clearances) for remifentanil to predict drug levels in lean and obese patients (Table 40-6).57

Using these weight-adjusted pharmacokinetic parameters, simulations of remifentanil effect site concentrations in obese and lean patients can be compared when dosed according to TBW versus LBM (Fig. 40-21). This set of simulations was performed for a lean woman (125 lb) and an obese woman (220 lb) of the same height (5 ft, 5 in). The simulations used dosing regimens typical for remifentanil as part of a combined technique with a potent inhaled agent, nitrous oxide, or a continuous infusion of a sedative hypnotic. The dosing scheme included a bolus dose (1 μg/kg) followed by a continuous infusion at 0.25 μg/kg/min for 20 minutes and then 0.15 μg/kg/min for 60 minutes.

Key points illustrated by these simulations include (1) Dosing obese patients according to their TBW can yield remifentanil effect site concentrations that are substantially higher than equivalently dosed lean patients. The peak effect site concentrations after 20 minutes of the anesthetic were 9.5 and 5.4 ng/mL for the obese and lean patient, respectively. (2) With the dosing scaled to LBM, the resultant simulated effect site concentrations for both the lean and obese patients were lower than the effect site profile for a lean patient dosed to her TBW. These simulations suggest that dosing IV anesthetics to LBM, similar to IBW, may lead to significant underdosing.

LBM has an additional limitation. When a patient's weight approaches morbid obesity, dosing according to the LBM becomes increasingly inaccurate.62 A unique and somewhat worrisome feature to the LBM equation is that for any given height, there is a maximum LBM. For TBWs that are above the maximum LBM, the LBM decreases. In Fig. 40-22, an LBM normogram, the LBM estimates that correspond to TBWs in which the LBM decreases with increasing TBW are plotted as a gray dashed line. This is most likely an artifact of the LBM equation. If not accounted for, very large patients would receive smaller doses than patients of equivalent height who are not as large. Another limitation of using the LBM normogram for dosing is that the LBM reports the patient mass in the absence of any fat tissue.

The LBM is always lower than the IBW. This is illustrated in the normogram with solid gray lines. These lines point out the resultant LBM for an IBW for a man who is 6 ft, 1 in tall and a woman who is 5 ft, 9 in tall. Because even lean people have some fat, this suggests that using the LBM for dosing anesthetics, as observed in Fig. 40-21, leads to some degree of underdosing. Hence, for clinical purposes, the LBM equation is perhaps only useful until reaching the maximum LBM for a given height with the expectation that many patients may require additional anesthetic.

With this limitation in mind, Bouillon and Shafer62 recommended that estimates of the ideal dosing weight from which to formulate dosing regimen use a modified normogram of LBM. Their modified normogram uses the following guidelines: (1) Conventional LBM scales are adjusted to IBW (in other words, for a patient who is at his or her IBW, the LBM at a given height should be the IBW). In effect, this shifts the original normograms presented in Fig. 40-22 up and to the left. (2) The scaled or modified LBM normograms are truncated at the maximal estimates of LBM. This approach relies on the assumptions that recommended doses printed in package inserts are appropriate for patients who are at their IBW. The scaled normogram of LBM is presented in Fig. 40-23. Again, the gray dashed lines represent the point where the scaled LBM decreases with increasing weight. Scaled LBM estimates in this region should not be used to estimate the proper dosing weight.

Using this scaled normogram for dosing remifentanil when a person's weight is at or below his or her IBW is unnecessary. Using the person's actual weight will suffice. When a person's weight is above his or her IBW, using the normogram in Fig. 40-23 may be useful but probably not immediately practical if not readily available in a clinical setting. A more simplified approach based on the normograms presented in Fig. 40-23 is to use a patient's IBW or if approaching morbid obesity, consider using their IBW plus some fraction of the difference between TBW and IBW.62 A simplified summary of scaled LBM normograms is presented by gender in Table 40-7. To account for height and weight, suggested scaled weights to use for dosing are presented as a function of the body mass index (BMI).

In Fig. 40-21, the green line represents the resultant remifentanil effect site concentration when dosing the 5-ft, 5-in (165-cm), 220-lb (100-kg) patient according to the suggested dosing weight in Table 40-6. Her IBW is 57 kg. Her BMI was calculated as 37. Twenty percent of the difference between the TBW and IBW is 8.6 kg. The suggested dosing weight is therefore 65.6 kg. The resultant effect site concentrations are in between the concentrations when dosing her according to her TBW and LBM, respectively. Using this approach, the inherent risks of underdosing using the LBM and overdosing using the TBW as well as avoiding the limitations of dosing according to the IBW are avoided.

Age and Intravenous Anesthetic Pharmacology: Developing Rational Dosing Strategies

Considering the aging of the world's population, the clinical significance of age in the practice of anesthesia is obvious. Practitioners are frequently faced with the clinical challenge of anesthetizing older adults, sometimes even patients in their ninth or tenth decades of life. Clinicians have long recognized that elderly patients usually require smaller dosages of most IV anesthetics to produce the desired therapeutic effect while minimizing adverse effects.

Of the frequently considered patient covariates (eg, gender, age, body weight, kidney function, hepatic function), age is perhaps one of the most valuable in terms of developing a therapeutic, nontoxic dosage strategy for many IV anesthetics. Age is easily measured (ie, just ask the patient), and its influence on the pharmacokinetics and pharmacodynamics of many IV anesthetics has been described in some detail.

With regard to age, both remifentanil and propofol can serve as prototypes to examine the impact of age on anesthetic pharmacology. Studies specifically designed to assess the influence of age on remifentanil and propofol pharmacokinetics and pharmacodynamics have been conducted, and complex models have been constructed that characterize the influence of age in quantitative terms.11,32,60,62,63

With regard to remifentanil, it is clear that the elderly do indeed require less drug to produce the desired spectrum of opioid effects. The reduced dosage requirement is a function of both pharmacokinetic and pharmacodynamic mechanisms, although pharmacodynamic factors dominate.11 The concentration of remifentanil necessary for 50% of maximal effect as measured by the EEG is markedly decreased in elderly adults (C50 for EEG changes); in addition, remifentanil's clearance and volume of distribution are also decreased.11 These age-related changes translate clinically into the need for a very substantial dosage reduction in elderly adults that is based largely on the increased potency of remifentanil in older patients (ie, a pharmacodynamic difference).

Using the pharmacokinetic–pharmacodynamic parameters from age-adjusted pharmacologic models of propofol and remifentanil, computer simulations can be performed to explore the suitability of various dosage schemes. Fig. 40-24 presents a set of simulations of propofol's pharmacologic behavior after an induction dose of propofol. Using age-adjusted pharmacokinetic and pharmacodynamic parameters,32,60 4 simulations of a 2-mg/kg propofol bolus for a 20-, 40-, 60-, and 80-year-old patient demonstrate that with the same dose, there is a subtle rise in the peak propofol effect site concentration and a small delay in reaching the peak with increasing age. By contrast, the concentration at which there is a 50% probability of losing unresponsiveness (C50 for unresponsiveness) drops by 50% from age 20 to 80 years (Table 40-8). Combining the pharmacokinetic and pharmacodynamic models, the duration of propofol effect (ie, the time above the effect site concentration associated with a loss of responsiveness) substantially increases with increasing age.

Similarly, Fig. 40-25 is an illustration of simulations of remifentanil's pharmacologic behavior after a 1-hour infusion. Using age-adjusted pharmacokinetic and pharmacodynamic parameters,11,63 4 simulations are presented of a 1-hour continuous infusion at a rate of 0.2 μg/kg/min for a 20-, 40-, 60-, and 80-year-old patient. With the same infusion rate, the remifentanil effect site concentration develops a substantial increase over time with increasing age (Table 40-8). Estimates of the C50 for analgesia allow for the prediction of duration of analgesic effect after termination of the 1-hour infusion. In Fig. 40-25, the C50 for each simulated age is presented as a horizontal line. After termination of the infusion, the duration of analgesic effect markedly increases with age (>3-fold increase from age 20-80 years).

Additional simulations can be used to estimate the percent reduction in dose as a function of age for both propofol and remifentanil. As is illustrated in Table 40-8, achieving equipotent doses in 20- and 80-year-old adults requires that an 80-year-old patient receive a dose that has been reduced by 55% to 65% of that which would be given to a 20-year-old patient. These simulations emphasize the importance of considering age when formulating an appropriate dose in elderly patients.

Although physiologic mechanism of the pharmacodynamic changes in elderly people remains largely unexplored, the pharmacokinetic changes may be at least attributable to decreased cardiac output. The lower cardiac output associated with advanced age64 presumably results in slower drug mixing and therefore higher peak concentrations after a bolus dose.65-67 Lower cardiac output may also decrease drug delivery to metabolic organs, resulting in lower clearance for some drugs.

When generalizing to other drugs, whether reduced cardiac output is the primary underlying mechanism responsible for the pharmacokinetic changes observed in elderly adults, it is consistent with the observation that many IV anesthetics (thiopental, propofol, etomidate) appear to have a smaller distribution volume or slower clearance in elderly adults.10,60,68,69

It is, however, important to point out that reduced cardiac output is not a ubiquitous finding in elderly adults, particularly in the absence of heart disease in well-conditioned individuals.70 Recognizing this has perhaps led to the common clinical notion of identifying a patient's "physiological" age instead of relying on chronological age alone.71,72 Significant reductions in dosage may not therefore be necessary for physically robust elderly patients with normal body habitus and without substantial coexisting disease.

It is more difficult to generalize with regard to age-induced pharmacodynamic changes to other IV anesthetics. Although elderly adults clearly have a "left-shifted" concentration–effect relationship for opioids (ie, the opioids are more potent in elderly patients11,73), a good deal of data suggest that older patients are not more pharmacodynamically sensitive to the sedative–hypnotics. For example, there is no difference between old and young in terms of the EEG C50 for etomidate or thiopental.10,68,69 On the other hand, recently published data suggest that both propofol and midazolam are more potent in elderly patients.32,74 Thus, although there is certainly general consensus that elderly patients require less medication than younger patients, whether this reduced dosage requirement can be attributed to pharmacokinetic or pharmacodynamic mechanisms remains unclear for some individual agents.

Blood Volume and Intravenous Anesthetic Pharmacology: Developing Rational Dosing Strategies

Dr Halford, a surgeon, wrote a letter to the editor of Anesthesiology after caring for several trauma victims after the attack on Pearl Harbor in 1941. He noticed that anesthetists had started using the IV anesthetic sodium pentothal. His comments included: "Then let it be said that intravenous anesthesia is also an ideal form of euthanasia… .With this heterogeneous mass of emergency anesthetists, it is necessary to choose an anesthetic involving the WIDEST MARGIN OF SAFETY for the patient… . Stick with ETHER."75

Anesthesiologists have long recognized the need to select certain IV anesthetics over others, to incrementally dose these anesthetics, and to moderate the overall dose for patients who have significant blood loss before or during surgery. Through experience, clinicians have learned that a full dose of selected IV anesthetics can lead to pronounced and often unwanted side effects with potentially disastrous consequences.

In the recent past, several researchers have attempted to quantify how the extent of blood loss impacts IV anesthetic pharmacokinetics and pharmacodynamics to include work with opioids,76-78 sedative–hypnotics,79-83 benzodiazepines,84,85 and local anesthetics.86 The most important finding consistent throughout this body of work is that equivalent dosing leads to higher drug concentrations with severe blood loss compared with control participants without bleeding. In addition, although derived volumes and clearances from pharmacokinetic analyses do not reflect true organ drug distribution and clearance, they do indicate that in severe blood loss, blood flow to muscle, gut, liver, and connective tissue is markedly decreased such that anesthetics delivered IV are most likely pumped straight to the brain in higher concentrations. This phenomenon leads to higher brain concentrations of anesthetic drugs and a more pronounced or prolonged anesthetic effect.87

As an example, Fig. 40-26 illustrates the differences in blood concentrations that result from identical doses of remifentanil in bled and unbled swine.78 After severe blood loss (42 mL/kg), resultant remifentanil blood levels were 2-fold higher during and after a 10-minute remifentanil infusion. Of note in this study, the dose (10 μg/kg/min) is approximately 50- to 100-fold more than a typical dose of 0.1 to 0.2 μg/kg/min, yet all animals survived despite losing more than half of their blood. A decrease in blood volume and cardiac index (5-1.7 L/min/m2) along with compensatory changes in regional blood flow are the likely physiologic mechanisms explaining these pharmacokinetic changes. Pharmacokinetic analyses revealed that the volumes of distribution and clearances were decreased in bled animals compared with unbled control subjects. Spectral edge changes in the EEG were used to measure drug effect. By contrast to the pharmacokinetic analysis, there was no difference observed in the pharmacodynamics between groups. As has been observed with fentanyl77 and morphine,76 these findings with remifentanil corroborate the relative forgiving posture of high-dose opioids on cardiovascular function even when used in hemorrhagic shock.

In contrast to opioids, blood loss has a more worrisome impact on propofol pharmacokinetics and pharmacodynamics. Similar to the experimental design used with remifentanil described above, investigators have bled animals and then administered propofol. Two major differences were observed when compared with remifentanil. First, the administration of propofol after severe hemorrhage (42 mL/kg) led to certain cardiovascular collapse. Second, the dose found to elicit a pharmacologic effect (ie, a change in the BIS of at least 50) in unbled animals was in no way tolerated in bled animals. To conduct experiments in bled animals, the propofol dose had to be reduced by more than 50%, and the extent of hemorrhage had be markedly reduced to 30 mL/kg. In this case, hemorrhage led to a decrease in the cardiac index from 5 to 2.6 L/min/m2. Subsequently, animals received a 10-minute propofol infusion at 200 μg/kg/min. Of interest, with equivalent dosing, the hemorrhaged animals exhibited approximately 2-fold greater plasma concentrations of propofol throughout the study period. A pharmacokinetic analysis revealed that similar to remifentanil, propofol compartmental clearances and volumes were decreased in bled animals.

The BIS was used as a surrogate measure of propofol effect. As expected, changes in the BIS lagged behind the changes in plasma propofol concentrations. These data were used to estimate the keo for bled and unbled animals and construct previously described pharmacodynamic models to include estimates of the C50 and γ. Comparison of pharmacodynamic parameters between bled and unbled animals revealed a similar keo and γ but a 2.7-fold decrease in the C50 (4.6 μg/mL vs 1.7 μg/mL for the control and shock groups, respectively). This is emphasized by the leftward shift (green arrow) in the C50 for each study group in Fig. 40-27.

Perhaps one of the more dangerous uses of IV anesthetics is during the induction of anesthesia. Here bolus doses are used to rapidly render a patient analgesic or unconscious, but these doses can be associated with significant morbidity if dosing does not account for large changes in blood volume. For purposes of discussion, consider an induction dose of propofol. The pharmacokinetic and pharmacodynamic findings related to propofol during blood loss described previously will be used to illustrate how blood loss impacts conventional dosing of sedative hypnotics.

In Fig. 40-11, using pharmacokinetic parameters previously described in humans,3 the onset and duration of loss of responsiveness after a propofol bolus of 2 mg/kg were 1 and 8 minutes, respectively, assuming that the necessary propofol effect site concentrations required for loss of responsiveness near 1.8 μg/mL.25 Conducting a propofol bolus simulation after moderate blood loss (35% of the blood volume) yields a significantly different result (Fig. 40-28). Based on the impact of blood loss on propofol pharmacokinetics and pharmacodynamics, this simulation takes in account the 2.5-fold increase in effect site propofol concentration and the 2.7-fold decrease in concentration required for loss of responsiveness.

With this simulation, the impact of moderate blood loss on the duration of effect is easily appreciated. Of note, with blood loss, there is more than a 5-fold increase in the duration of effect (8-44 minutes). These simulations suggest that propofol should be used, if at all, with extreme caution! Estimating the dose that would provide an equivalent effect in a person with severe blood loss compared with a person with normal cardiovascular physiology yields a propofol dose reduction of 80% (eg, 0.4 mg/kg). Although the impact of blood loss on propofol is dramatic, it is important to recognize that this simulation is of a single propofol bolus and does not reflect the common practice of combining propofol with an opioid during the induction of anesthesia. In this scenario, it is likely that the pronounced increase in peak effect and duration of effect would only be larger and potentially more dangerous.

Perhaps the most important consequence of blood loss on propofol behavior is the exaggerated hemodynamic response after a bolus dose. Propofol is a peripheral vasodilator and suppresses contractility.88 As observed in these simulations, a propofol bolus dose yields higher effect site concentrations that remain elevated for a prolonged period of time, thus amplifying propofol's cardiovascular depression. Sodium thiopental has a similar profile of cardiovascular depression, and this likely explains why Dr Halford was so adamant about the dangers associated with the induction of anesthesia with sodium pentothal.

What about resuscitation? Does volume resuscitation restore drug disposition and effect to baseline? In typical clinical practice, some fluid resuscitation is usually under way before the administration of an anesthetic. Based on the premise that resuscitation will restore cardiac output and systemic blood flow, the shock-induced pharmacokinetic and pharmacodynamic changes may be reversed. In a similar set of experiments, a comparison was made between unbled control subjects and bled and then partially resuscitated swine.89 Hemorrhage was severe (42 mL/kg), and resuscitation constituted an infusion of crystalloid to maintain a mean arterial blood pressure of 70 mm Hg for 60 minutes. This resulted in a resuscitation volume of 59 mL/kg.

After a 10-minute high-dose (750 μg/kg/min) propofol infusion, the propofol plasma concentrations were nearly identical. Resuscitation restored the shock-induced changes in propofol pharmacokinetics to near baseline values. The pharmacodynamic parameters, however, remained altered. As with severe blood loss, the C50 was decreased 1.5-fold after hemorrhage and resuscitation. Although the mechanism for this phenomenon is not well understood, one explanation for the increase in end-organ sensitivity to propofol may be at least partly attributable to an unrecognized increase in unbound propofol. Thus, the leftward shift in the C50 of propofol may represent an undetected pharmacokinetic difference between groups. Although the plasma propofol levels were similar between the bled and then resuscitated animals and the unbled animals, the amount of unbound propofol available to exert a pharmacologic effect may have been increased.90 After removing more than 50% of the estimated blood volume and replacing it with crystalloid, plasma protein content would most likely be decreased. Furthermore, alterations in organ blood flow, capillary wall integrity, and plasma pH may influence the levels of unbound propofol. Given that plasma protein content, propofol-plasma protein binding, or unbound propofol levels were not measured or estimated, the extent that changes in unbound propofol played in altering the observed differences in end-organ sensitivity remains unknown.

With this protocol, 60% of the estimated blood volume was removed. A total of 140% of the shed blood volume was replaced with lactated Ringer solution to maintain a near-normotensive blood pressure. The near-normal blood pressure was deceiving! Although hemodynamic function appeared near normal (ie, the central venous pressure and cardiac index were similar to those in unbled animals), the cardiovascular response to propofol remained exaggerated. During the propofol infusion, the cardiac index decreased by 1.7 L/min/m2 in the shock-resuscitation group but only by 0.2 L/min/m2 in the control group. The large hemodynamic changes in the shock-resuscitation group illustrate how severe blood loss followed by partial resuscitation can lead to potentially large cardiovascular changes with the administration of propofol. In fact, a significant clinical correlate from this analysis is that despite a near-normal hemodynamic profile after partial resuscitation for severe blood loss, resuscitation should continue to minimize the potentially severe hemodynamic depression that can be associated with the administration of propofol.

Fig. 40-29 represents a simulation of the propofol effect site concentration after a propofol bolus dose in a patient with severe blood loss followed by partial resuscitation with crystalloid (1.5 mL of crystalloid per 1 mL of estimated blood loss). This simulation accounts for the pharmacodynamic changes as manifest by a 1.5-fold decrease in the effect site concentration required for loss of responsiveness. The duration of effect increases from 8 to 11 minutes.

Compared with propofol, both ketamine and etomidate have greater acceptance among clinicians who care for patients with significant blood loss. This is largely because the cardiovascular depression known to be exaggerated with propofol and sodium pentothal is not as apparent with etomidate and even to a lesser extent with ketamine. Although etomidate is known to produce mild cardiovascular depression, prior work surprisingly has revealed minimal cardiovascular change after a high-dose brief continuous etomidate infusion81,83 during moderate hemorrhagic shock (30 mL/kg). In a similar fashion, the pharmacokinetic and pharmacodynamic profile of etomidate after blood loss was also minimally influenced by blood loss. This suggests that dosing requirements for etomidate do not require adjustment after moderate blood loss. This finding goes along with the widely held view that etomidate is a good choice in hemodynamically unstable patients.

What remains unknown is the impact of blood loss on ketamine. Preliminary work has revealed that, similar to etomidate, blood loss does not significantly impact the pharmacokinetics of ketamine. Ketamine is known to increase sympathetic tone, serve as a potent analgesic, and perform favorably in patients with poor cardiovascular function. These preliminary findings support the widely held view that ketamine is an important drug to maintain in our pharmacologic armamentarium when caring for patients with life-threatening blood loss.

The pharmacodynamic properties of ketamine, however, are difficult to assess. This is largely because ketamine is a racemic mixture that, when metabolized, has an active metabolite, norketamine. Hence, the contribution of both enantiomers and norketamine must be considered when assessing the overall drug effect of ketamine. An additional difficulty with measuring ketamine's drug effect is that it is difficult to identify surrogate measure for ketamine's effect. For example, the BIS is not a reliable measure of ketamine's sedative effects.

In summary, as clinicians manage patients with blood loss through often perilous anesthetics, hemorrhage and even hemorrhage followed by resuscitation that appears to restore hemodynamic function to near normal can lead to dramatic alterations in the pharmacologic behavior of commonly used sedative hypnotics and opioids. Duration of effect, peak effect site concentrations, and extent of cardiovascular depression should all be considered when selecting an IV anesthetic and formulating an appropriate dose. Hemodynamically compromised patients are especially susceptible to the cardiovascular suppression of selected sedative hypnotics, but other sedative hypnotics appear to be much safer. Propofol and sodium pentothal are especially poor choices even after some degree of resuscitation. By contrast, ketamine and etomidate tend to be immune to the deleterious effects of moderate to severe blood loss on their pharmacokinetic profiles. Severe blood loss alters opioid pharmacokinetics, leading to higher plasma concentrations, but opioids, in contrast to propofol and sodium pentothal, enjoy a wider therapeutic margin in the presence of blood loss. What remains unexplored, however, is the impact blood loss has on the resultant effect from the simultaneous administration of multiple drugs. As illustrated in the drug synergism section of this chapter, opioids and sedative–hypnotics can dramatically influence one another. How this interaction behaves in the presence of intravascular volume depletion remains unknown.

Opioid-Tolerant Patients

One of the most vexing problems facing anesthetists is satisfactorily caring for patients who chronically consume opioids (see Chapter 24 for a general review of substance abuse and anesthesia practice). With chronic consumption of opioids, tolerance develops, and more drug is required to achieve a desired analgesic effect. This becomes especially problematic in the perioperative period when opioid dosing requirements for these patients often reach thresholds associated with significant morbidity in the nonopioid-tolerant population. Out of concern for patient safety, opioid doses are often administered in more conventional doses to avoid unwanted side effects associated with opioid toxicity, leading to often dramatically poor pain control in the postoperative period. In these types of patients, the population-based pharmacodynamic models used to describe opioid concentration–effect relationships become obsolete. Performing simulations to estimate opioid behavior either alone or in combination with other types of anesthetics using pharmacodynamic models built from studies evaluating an opioid-naïve population is bound to provide a faulty estimate of drug effect in patients who chronically consume opioids. The essential problem is that the magnitude of the right shift in the C50 and C95, which are key parameters used to build the pharmacodynamic models, is unknown (Fig. 40-30).

With the widespread use of oral short-acting and time-contingent opioids, transdermal opioid delivery systems, and implantable opioid infusion pumps, opioid tolerance is recognized as a growing challenge in the perioperative environment. Nevertheless, there is a paucity of literature examining the phenomenon of opioid tolerance as a covariate in pharmacodynamic models. One potential reason for this is that opioid tolerance is a difficult feature of drug behavior to quantify and most likely varies substantially from person to person based on duration of opioid consumption and opioid dose.

With this problem in mind, investigators have explored methods of identifying the concentration—effect relationship for opioid on an individual basis.91-93 Using principles of pharmacokinetics and pharmacodynamics, a technique has been developed called the fentanyl challenge. The fentanyl challenge was designed for use in patients with a known history of chronic opioid consumption who are scheduled to undergo surgical procedures associated with significant postoperative pain that will require a general anesthetic and are of moderate to long duration (ie, multilevel lumbar spine instrumentation). The fentanyl challenge protocol calls for a rapid continuous infusion of fentanyl (2 μg/kg/min) to be administered until the onset of respiratory depression in the absence of any other sedatives, anxiolytics, or opioids (Fig. 40-31). The optimal setting in which to perform the challenge is just before the induction of general anesthesia. The onset of respiratory depression as defined within the fentanyl challenge is a respiratory rate less than 6 breaths/min. A critical component of the challenge is to measure the time from the onset of the fentanyl infusion until the onset of respiratory depression.

With the duration of the high-dose fentanyl infusion until respiratory depression identified, a pharmacokinetic simulation of the infusion is made to identify the fentanyl effect site concentration that is associated with respiratory depression. This process redefines the pharmacodynamic relationship between fentanyl effect site concentrations and drug effect that is unique to a particular patient. A major assumption of this protocol is that more drug will be required to achieve analgesia.

As previously described in the pharmacodynamic section, there are a series of pharmacodynamic relationships over an array of noxious stimuli encountered in the perioperative environment (Fig. 40-10). One interesting feature of this array of concentration relationships is that they can be quantified in terms of a percentage of the amount of drug required to elicit EEG changes.26 For example, the fentanyl effect site concentration associated with C50 for changes in the spectral edge, a measure of EEG activity is approximately 9 ng/mL.35,73,94,95 The fentanyl effect site concentration C50s associated with analgesia and respiratory depression are 1.6 and 5.4 ng/mL, respectively. Hence, the fentanyl effect site concentrations for analgesia and respiratory depression are 17% and 60%, respectively, of the C50 for EEG changes. Taking advantage of this linearity, the analgesic fentanyl effect site concentrations are approximately 30% of those associated with respiratory depression.96,97

Although not well established, preliminary work has indicated that the linear relationship between analgesia, respiratory depression, and EEG changes remains intact despite a rightward shit in the concentration effect relationships associated with opioids in chronic opioid-consuming patients.91-93 Using this linearity, the fentanyl challenge is able to predict analgesic effect site concentrations from estimated concentrations of fentanyl required to produce respiratory depression. For example, the fentanyl effect site concentration associated with analgesia is 30% of the effect site concentration associated with respiratory depression regardless of whether the fentanyl effect site concentration is required to achieve respiratory depression.

With the analgesic effect site concentration identified as a percentage of the concentration required for respiratory depression, the next step is to develop fentanyl dosing regimens to achieve and maintain analgesia. Dosing goals may be directed at providing intraoperative analgesia as part of a combined anesthetic technique or postoperative analgesia until the patient is able to take oral analgesics. Additional simulations are used to identify optimal infusion rates that will produce the target analgesic effect during the course of the anesthetic and into the postoperative period.

Unique features of fentanyl pharmacokinetics that are important to consider when initiating an infusion following a fentanyl challenge include

1. With prolonged continuous infusions (ie, >2 hours), effect site concentrations continue to rise. Simulations of continuous infusions reveal that effect site concentrations do not reach near steady state for at least 24 hours. Thus, it is important to anticipate the duration of a postoperative continuous fentanyl infusion and select an infusion rate that will not exceed the effect site concentrations associated with respiratory depression.

2. Upon completion of the fentanyl challenge before the induction of anesthesia, a significant amount of fentanyl can be delivered depending on the duration of the challenge. For example, an 8-minute infusion at 2 μg/kg/min in a 90-kg patient will result in the delivery of 1440 μg (29 mL) of fentanyl. This initial dose should be accounted for when formulating an intraoperative fentanyl infusion rate.

In Fig. 40-32, a set of simulations illustrate the fentanyl effect site concentrations that result from a fentanyl challenge. In this case, an 8-minute infusion was required to reach the onset of respiratory depression. After the challenge are 6 simulations of a fentanyl infusion ranging from 1 to 6 μg/kg/h for a 6-hour anesthetic. At 8 minutes, the fentanyl effect site concentration associated with respiratory depression is 20 ng/mL, giving a target effect site concentration of 6 ng/mL (ie, 30% of 20). Using this simulation, the fentanyl infusion that best approximates the target effect site concentration of 6 ng/mL is a continuous infusion at 3 μg/kg/h.

Information obtained from pharmacokinetic simulations after a fentanyl challenge can be used to improve intraoperative dosing of fentanyl to ensure adequate analgesia in the early postoperative period. In addition, this same information can be used to identify IV fentanyl dosing regimens for the first 24 to 48 hours after selected surgical procedures associated with significant postoperative pain and no or inadequate ability to use oral analgesics.

To improve intraoperative dosing of fentanyl, prior work has found that administering a continuous infusion of fentanyl that maintains the analgesic effect site concentration throughout the duration of a surgical procedure leads to an adequate level of analgesia in the early postoperative period yet allows for timely emergence from anesthesia in patients who chronically consume opioids.92 With regard to postoperative dosing of fentanyl, information gained from the fentanyl challenge can be used to identify dosing regimens for patient-controlled analgesia (PCA) combined with a basal continuous infusion to maintain analgesia while avoiding respiratory depression.91-93 Infusion rates (in μg/kg/h) used intraoperatively to target the fentanyl concentrations associated with analgesia are divided in half. Half the infusion rate is administered as a continuous infusion, and the other half is administered as intermittent boluses using a PCA.

Preliminary exploration into the efficacy and safety of this technique has been encouraging. In a cohort of patients who reported chronic consumption of opioids, a fentanyl challenge was used to identify the intraoperative and postoperative dosing regimens for fentanyl. The number of interval doses delivered via the PCA was used as a metric of pain control. Interval dose requirements of 2 doses or fewer per hour were considered to provide adequate analgesia. No use of the PCA over a 4-hour period was considered to be an aggressive basal infusion and was decreased by 20%. PCA usage more than twice an hour was considered in inadequate basal infusion and was increased by 20%. After 24 hours of this dosing regimen, measures of respiratory function, arterial PCO2 levels, and pain control were made. In all subjects, respiratory rates and blood oxygenation were normal. Arterial PCO2 levels ranged from 40 to 47 mm Hg. PCA usage was within 1 to 2 doses per hour.

A computer simulation of a fentanyl challenge followed by the intraoperative and postoperative course is presented in Fig. 40-33. In this example, an 80-kg patient known to chronically consume opioids required 10 minutes to achieve the onset of respiratory depression. The corresponding fentanyl effect site concentration at this time was 27 ng/mL. The target effect site concentration for analgesia was therefore 30% of 27 or approximately 8 ng/mL. Subsequently, a basal infusion of fentanyl was administered during the 4-hour intraoperative period at 5 μg/kg/h as part of a combined technique with a potent inhaled agent. Upon completion of the intraoperative period, the patient was allowed to emerge from anesthesia. In the postoperative phase, a basal infusion was started at 2.5 μg/kg/h. In addition to the basal infusion, the PCA was set to deliver a demand dose of 50 μg every 15 minutes (2.5 μg/kg/h if using all 4 doses). The basal infusion and PCA were used for 36 hours after the anesthetic. The average PCA usage was 2 demand dose per hour. No adjustments were made in the PCA.

Several points of clinical interest are illustrated by this simulation. By starting a fentanyl infusion rate at 5 μg/kg/h immediately after the fentanyl challenge, the resultant effect site concentrations were well above the target concentration for nearly 90 minutes. This is an important feature to consider when delivering anesthetics of shorter duration. Also of interest, by the end of the 4-hour anesthetic, the effect site concentration was beginning to climb above 8 ng/mL. If an anesthetic requires more time, perhaps a more moderate infusion rate would be prudent (ie, 4 μg/kg/h). In the postoperative phase, the simulation reveals that the basal infusion rate in combination with the PCA maintained the target concentration well. On being turned off, the fentanyl effect site concentrations drop fairly slowly. The time required for the fentanyl effect site levels to drop by half is more than 5 hours. This dissipation time may be important to consider when initiating oral analgesic therapy after terminating a PCA and basal infusion.

• Bouillon T, Shafer SL. Does size matter? Anesthesiology. 1998;89(3):557-560.
• Davis JJ, Swenson JD, Hall RH, et al. Preoperative "fentanyl challenge" as a tool to estimate postoperative opioid dosing in chronic opioid-consuming patients. Anesth Analg. 2005;101(2):389-395; table of contents.
• Egan TD, Shafer SL. Target-controlled infusions for intravenous anesthetics: surfing USA not! Anesthesiology. 2003;99(5): 1039-1041.
• Fisher DM. (Almost) everything you learned about pharmacokinetics was (somewhat) wrong! Anesth Analg. 1996;83(5):901-903.
• Krejcie TC, Avram MJ. What determines anesthetic induction dose? It's the front-end kinetics, doctor! Anesth Analg. 1999;89(3):541-544.
• Minto CF, Schnider TW, Egan TD, et al. Influence of age and gender on the pharmacokinetics and pharmacodynamics of remifentanil. I. Model development. Anesthesiology. 1997;86(1):10-23.
• Minto CF, Schnider TW, Short TG, et al. Response surface model for anesthetic drug interactions. Anesthesiology. 2000;92(6): 1603-1616.
• Shafer SL. Shock values. Anesthesiology. 2004;101(3):567-568.
• Shafer SL, Stanski DR. Improving the clinical utility of anesthetic drug pharmacokinetics. Anesthesiology. 1992;76(3):327-330.
• Sheiner LB, Stanski DR, Vozeh S, et al. Simultaneous modeling of pharmacokinetics and pharmacodynamics: application to d-tubocurarine. Clin Pharmacol Ther. 1979;25(3):358-371.