The uptake and action of anesthetic in the brain are dynamic processes that are clinically important. The standard measures of “depth” of anesthesia use a single feature to define a discrete point on what is probably a continuous dose-effect relationship. For example, the presence or absence of movement after a standard stimulus1 is used to define the standard of anesthetic potency, minimum alveolar concentration. An alternative, continuous measure of anesthetic activity is obtained from the processed electroencephalogram (EEG), which gives a value, quantified using an arbitrary scale with set limits, to indicate depth of anesthesia. This measure has advantages compared with the individual “set points” of categorical responses. For example, the onset of anesthetic effect can be followed as drug concentration increases, so that the kinetics of drug uptake can be assessed continuously.2 However EEG signal processing requires time for signal acquisition and analysis, so that EEG measures lag behind changes in brain activity.3,4
Another way to obtain information about the onset and offset of drug action is to use a graded biological response to directly measure drug action. This is often done with other drugs, such as analgesics, when the subject provides a measure of the drug effect. We gave subanesthetic doses of anesthetic drugs to volunteers and showed dose relationships for different subjective and objective effects for nitrous oxide, sevoflurane, and alcohol.5 However tests of this type take time to perform, so the time course of rapid events such as the onset of anesthesia cannot be measured accurately. If we had a test of drug effect that gave nearly immediate results, then we could measure the progress of onset of the drug effect.
We measured effects of nitrous oxide on central nervous system (CNS) function, during onset and offset, using such a test, which gives frequent estimates of brain function. We related these measures of drug effect to predictions of CNS levels of nitrous oxide, estimated using end-tidal concentrations and a simple model of brain kinetics. We evaluated the relationship between dose and drug effect by plotting change in test performance in relation to brain concentrations that had been estimated using different kinetic settings for the model. We argue that with the correct kinetic model the plots during onset and offset will coincide, showing the unique relationship between drug effect and estimated brain concentration provided by that particular model setting.
The study was approved by the Lothian Research Ethics Committee. We recruited healthy volunteer subjects who were not taking any centrally acting drugs such as antidepressants or opioids. The subjects gave written consent. Female subjects were tested within the first 2 weeks of their menstrual cycle. Participants were asked not to consume any alcohol in the 12 hours before the study session and to get a good night's sleep. Before testing the subjects had a light breakfast or lunch and were allowed their usual caffeine intake until 4 hours before testing. Clear fluids were allowed up to 2 hours before testing.
Each subject was studied in a single session, which consisted of a short practice run, followed by three 32-minute test episodes. During the practice run, the subjects practiced the tapping test twice with the facemask off and then twice after the mask had been applied. A rest period of approximately 10 minutes was given after the practice run. Each test episode consisted of an initial 7-minute run-in period, a 15-minute period of test gas administration and then a 10-minute washout period (Fig. 1A). A 10-minute rest period was allowed after each test episode. Each subject then completed 3 test episodes. In each episode, the test gas used was different (Fig. 1B). The order of gas administration for the subject was determined by random allocation using sealed envelopes with a Latin square design. The test gases were 30% oxygen, balance nitrogen; 5% nitrous oxide, 30% oxygen, balance nitrogen; or 30% nitrous oxide, 30% oxygen, balance nitrogen.
We used an alternate tapping test and a digit symbol substitution test (DSST). For the alternate tapping test, the subject held a metal stylus and was instructed to tap alternately on 2 metal target plates, 10 cm square, fixed 50 cm apart. The subject was encouraged to tap as accurately and rapidly as possible for exactly 30 seconds. Stylus contact with each target generated a specific signal voltage that was recorded through an A/D converter (Micro 1401, CED, Cambridge, UK). The tapping rate was measured using commercially available software (Spike 2 version 3.01) and a personal computer. For analysis, data were exported to Excel (Microsoft, Seattle, WA) and graphical and statistical software (Prism, version 5.00, GraphPad Software, San Diego, CA).
The DSST used a set of response sheets that we used previously.6 At the top of a sheet of paper, a key is shown that matches 9 single digits, 1 to 9, with simple symbols. Below this reference key is a table of boxes. Each box contains a random single digit above and a space below. The subject was handed this sheet of paper at the appropriate time in the test run. The subject was then signaled to start and had to write the corresponding symbol in the space below each successive number as rapidly as possible. The first 10 symbol substitutions are not scored and thus allow practice to familiarize the subject with that symbol set. After these first 10 boxes are completed, the actual measurement time starts as the subject continues to write substitute symbols in each successive box. The number completed in the following 90 seconds is recorded. A new sheet, with a different version of the digit-symbol key and a new set of random numbers, is used for each test.
During the test episodes, the subject breathed from a well-fitting silicone rubber facemask and a 2-way nonrebreathing valve (Hans Rudolph Inc, Kansas City, MO) connected to a T-Tube reservoir system. The exhaust gases were actively scavenged. Gases were supplied from an anesthetic machine (Aestiva5, GE Healthcare, Hatfield, Herts, UK). Gas was sampled continuously from the mask to measure oxygen, carbon dioxide, and nitrous oxide using a respiratory gas analyzer (Datex CD2 to 02 NormocapR 200, Datex Instrumentarium Corp., Helsinki, Finland). The carbon dioxide and nitrous oxide signals were recorded with the A/D converter (Micro 1401, CED, Cambridge, UK). End-tidal values of nitrous oxide were taken as the average of the last 50 milliseconds of expiration of each breath, using Spike software triggered from the carbon dioxide trace. The time at the end of the 50 milliseconds was noted. Values caused by swallowing were excluded. During the study, an independent investigator set the gas mixture composition and the machine controls and gas analyzer display were concealed from the subject and from the investigator administering the tests. Neither the investigator administering the tests nor the subject was aware of the identity of the gas, and the subject was not told when the gas composition was altered. After all 3 test runs were completed, the subject was asked to decide which run had been the 0%, 5%, and 30% nitrous oxide administration.
Data Analysis and Statistics
The end-breath times and end-tidal nitrous oxide concentrations from the 3 test periods were combined into a continuous trace, to model any possible carryover among the 3 test periods. There were 2 gaps of 10 minutes in respiratory data between runs (Fig. 1B). To fill these 2 gaps, end-tidal nitrous oxide values were interpolated. For each subject, we took the mean end-tidal nitrous oxide value and mean respiratory frequency for the final 30 seconds of measurements before the gap and for the first 30 seconds after the gap, when measurements started again. From these values we linearly interpolated a set of end-tidal values and times to fill the gap between each session (Fig. 1B). The entire dataset was then used to model brain concentrations of nitrous oxide.
We assume that end-tidal nitrous oxide values are equilibrated with arterial blood. We wished to model a partial pressure at the site where we considered nitrous oxide was exerting its effect and call this “brain” concentration. We used a washin type of model where arterial blood supplies a single tissue compartment, where there is instantaneous equilibration between the blood and tissue. The “concentration” values derived in the model indicate the partial pressure that would exist in the brain if it were equilibrated with gas of that partial pressure.
Using successive end-breath times and end-tidal nitrous oxide values, we estimated the brain concentration of nitrous oxide (Ce) by relating the rate of change of brain concentration to the difference between blood (i.e., end-tidal) and brain concentrations:
where keq is the rate constant that characterizes the speed of equilibration between the end-tidal gas and the cerebral tissue. The keq can be related to the equilibration half-time, which we use to quantify the equilibration rate7:
Equilibration between blood and tissue, indicated by the rate constant keq or the equilibration half-time, can be related to the relative perfusion of the tissue and the tissue/blood partition coefficient with the following equation:
The partition coefficient, or relative solubility of nitrous oxide in blood and active brain tissue, has a value that is close to unity. The usual units for tissue bloodflow are mL blood/minute per 100 g tissue, but because the density of brain tissue is close to 1 g/mL we may consider bloodflow to be the same if the units were changed to mL blood/minute per 100 mL of brain tissue. Thus an equilibration half-time of 2 minutes would indicate a perfusion rate of ∼34 mL/min/100 g tissue, and an equilibration half-time of 1 minute indicates 2-fold greater perfusion, ∼69 mL/min/100 g tissue.
Values for Ce were calculated using a Visual Basic function for Microsoft Excel obtained from PK-PD tools.a The function uses log-linear interpolation and numerical convolution to calculate values of Ce based on the measured end-tidal values and a given value of t½keq. We used t½ keq values of 0.5, 0.75, 1, 1.5, 2, and 3 minutes. These values cover the range of previous estimates of cerebral equilibration half-times, using similar models.8 The concentration values derived indicate the partial pressure that would exist in the brain if it were equilibrated with this concentration of nitrous oxide in a gas mixture at atmospheric pressure.
We used a repeated-measures ANOVA to assess how the gas affected the tapping test and the DSST. The DSST and tapping test results were tested for normal distribution with the D'Agostino and Pearson omnibus normality test. Factors were the order of gas administration and the concentration of nitrous oxide. Further analysis was with a post hoc multiple comparison test with Bonferroni adjustment because we had conducted 2 tests on the same sample, i.e., the tapping test and the DSST. We tested for interaction between order of administration and gas effect.
For each subject, the 3 control values before the gas administration started were compared to the last 3 tapping scores during nitrous oxide inhalation, using a 2-tailed unpaired t-test with Welch's correction. To assess the best fit of the modeled brain concentration to the effect on tapping speed, tapping data were expressed as a decrease from the control score. We calculated the absolute area of these plots (decrease in tapping speed versus modeled brain concentration) to indicate the relationship between the degree of looping of the plot and the half-time used to model the effect-site concentration. We used Prism v5.00 and SPSS v17.0 (SPSS, Chicago, IL) for statistical analysis and set P < 0.05 to indicate significance. Values given are mean (SD) unless otherwise stated.
We assessed 39 volunteers for eligibility. Four were excluded for medical reasons and 11 later declined to take part in the study. Of the remainder, 24 were admitted to the study, 6 male and 18 female, aged 21 (6) years, and weighing 70 (10)> kg. Four subjects were later excluded from analysis for technical reasons. In 2, mask fit was inadequate: 1 subject did not perform the tapping test correctly, and some data were inadvertently lost from 1 subject's records. A further subject received the gases in the incorrect order; these data were excluded from the initial statistical analysis of DSST responses (which included assessment of an order effect) but were included in all other analysis.
The mean end-tidal concentrations of nitrous oxide achieved during the 5% and 30% sessions were close to the target values, being 4.7% (0.5) and 27.2% (2.2), respectively. All subjects correctly identified 30% nitrous oxide, but only 61% were correct in judging which gas was 5% nitrous oxide.
The distributions of the DSST and tapping test values were normal. Nitrous oxide reduced the DSST (F = 8.73, P < 0.0001). Nitrous oxide, at 30%, reduced the DSST from 69.7 (10.4) to 53.5 (10.9) (P = 0.008) and DSST was not affected by 5% nitrous oxide. (Fig. 2) The order of gas administration did not significantly influence these effects (F = 2.35, P = 0.10) and there was no interaction between order of the gas administration and the effect of the gas (F = 0.54, P = 0.86).
The number of taps in a test period was reduced from 122 (15) to 113 (15) by 30% nitrous oxide (P = 0.018), but the effect was much more variable than for the DSST results, with some subjects showing an increase in tapping. In 9 of the 20 subjects, 30% nitrous oxide caused a significant reduction in tapping. In these subjects, tap number decreased from 127 (16) to 110 (17). Figure 3 compares the relative responses of DSST scores and tapping numbers in individual subjects, and also indicates those subjects in whom there was a significant reduction in the number of taps.
To relate the effects of nitrous oxide to the changes in tapping frequency, we modeled brain concentrations in those subjects who had a significant change in tapping frequency. Because maximum impairment to tapping frequency was reached within 5 minutes of starting nitrous oxide administration, we plotted the early data for washin and washout to relate the predicted brain concentrations to the drug effect. (Fig. 4) With a half-time of 2 minutes, the loop area was less than with other half-times.
We found a plausible fit between modeled concentration and the drug effect when a half-time of 2 minutes is used in this simple model. This is equivalent to a time constant of 0.34 minutes. If the blood/tissue partition coefficient for nitrous oxide in nervous tissue is approximately 1,9 then this would indicate that effect-site bloodflow is of the order of 34 mL 100 g−1 min−1. This is rather less than current measures of cerebral bloodflow10 and consistent with previous measures of cerebral bloodflow obtained during anesthesia with other drugs.7
The tapping test has been used by previous investigators6,11 and gives a convenient measure of subanesthetic impairment that can be easily and frequently repeated. The dose of nitrous oxide that we chose has significant effects, although previous reports only gave summary measures and had not indicated the individual susceptibility. Before the study was done, we did not appreciate that some subjects might be less affected than others, although with hindsight this was predictable. Although the test is a good measure of psychomotor impairment, it represents the outcome of several related processes, and we assume that the most susceptible of these is located in the nervous system. Although the task appears straightforward, many elements in task performance are involved, such as motivation, sensory input, central processing, coordination of movement, and motor activity. In addition factors such as learning, boredom, and distraction may be affected by anesthetics.12 Our study plan allowed us to sustain constant conditions, and the experiment was carefully designed to minimize or account for confounding effects. We believe that with this dose of inhaled drug, the most likely effect that we are detecting is a central one, which includes not only the cerebral cortex, but also subcortical and spinal systems involved with motor performance.
The process of “anesthesia” is not a single event at a unique site.13–15 Evidence shows that even simple reflex actions contain components with different susceptibility to anesthetics.16 Although nitrous oxide has well recognized effects on the EEG, these are not clearly evident when common methods of EEG analysis, such as Bispectral Index, are used.17,18 An alternative method of EEG processing has shown a more graded cortical effect during onset of nitrous oxide effects, attributed to a reduction in afferent activity.19
We have used a simple descriptive method to assess the data we obtained on the time course of the anesthetic effects. The effects we are assessing are small, and we cannot be certain that the dose-effect relationship being studied is linear. Previous studies with nitrous oxide at low doses disagree on this point. Fagan et al. found a linear relationship between reduced tapping frequency and nitrous oxide concentration, even at low doses,6 but others did not.20 We chose an analysis that did not require assumption of a linear response. We chose a robust grouped analysis. We decided not to attempt to determine the exact time constant that would provide a best fit for the data, and not to model individual responses using a standard dose-response relationship. Such an approach is not feasible when only a small segment of the dose-response curve is accessible for the fitting process, and uncertain if the SD of the measure of drug effect is as large as we found it.
We did not find any convincing impairment in 11 of 20 subjects using the tapping test. The DSST was more consistent, and impairment was present in all subjects. Clearly the DSST is more sensitive to the effects of 30% nitrous oxide. We do not believe that differences among subjects in susceptibility to the tapping test invalidate our conclusions. Differences in susceptibility are very likely. Animal studies have shown that nitrous oxide susceptibility is a stable genotype that can be accentuated by breeding,21 and anesthetic susceptibility varies with ethnic origin in humans.22 We believe that we have demonstrated that 30% nitrous oxide has a substantial effect on only half a sample of young subjects. A refinement of this study would have been to determine each subject's individual pharmacodynamic susceptibility and then use a dose based on this value to test onset. However this would have involved a considerably more complex experiment and it would have been unlikely to yield different estimates of the drug's kinetics.
Our measure of susceptibility was based on responses that were measured after central nervous equilibration was likely to be complete, so we selected subjects for the kinetic analysis solely on the basis of this specific pharmacodynamic characteristic. We have no reason to believe that those volunteers, who were more susceptible to nitrous oxide at steady-state, would be different for pharmacokinetic reasons. Indeed, if we included subjects in the analysis who were resistant to the effects of nitrous oxide, we would merely conceal the kinetic effects we wished to observe.
Finally, turning to kinetics, we used a very simple model of anesthetic transfer from the lung to site of action, which ignores the more complex components of this process.7 Our observations are consistent with a half-time of 2 minutes. Our model reduces the system to a single exponential mass transfer process, which can be considered to be a “washin” from arterial blood into a single tissue compartment. Other factors are “lumped” into this model, such as a difference between end-tidal and arterial partial pressure, lung to brain circulation time, and impaired diffusion,23 all of which would tend to prolong the half-time.
Olofsen and Dahan7 modeled anesthetic onset and offset kinetics, measuring the processed EEG. They used a sigmoid dose/effect relationship to relate the EEG effect to the calculated effect-site concentration. They were able to use this approach because a wider range of drug dose and effect was available to model. In the present study, only a small part of the dose/effect relationship could be examined, so the position and slope of the relationship cannot be easily fitted to a standard dose-response curve. Olofsen and Dahan7 calculated a theoretical t1/2 for sevoflurane of 2.1 minutes but found values between 2.3 and 3.5 minutes, estimates that have been supported by another study.24 The value they calculated is in fact compatible with our results, when the relative brain solubilities of sevoflurane and nitrous oxide are considered. Other authors have used more complex models with time factors, with variable results, but these were of injected drugs, where delays and mixing are more prominent features.25,26 When end-tidal measures of drug are available, a substantial part of the delay is avoided, in comparison with IV drugs.
One further feature of this study is that we measured a neural response directly in conscious subjects. The major delay in the process we studied is likely to be the time needed to wash in the drug in the relevant neural tissue site, but another reason for delayed neural responses may be changes in neural circuit kinetics, for example, in associations between cortical and subcortical activity. A study of the H reflex, which also provides a direct measure of anesthetic effect, found a greater half-time that could be either because of a different site of action, or different neuronal dynamics.24
A further reason for a delay in the time of onset of effect in some studies (but not the present one) may be the delay involved in computing the processed EEG signal, which can be of the order of 60 seconds, although this time varies depending on process used and the rate of change of the input signal.27 Even similar methods of EEG processing can yield different estimates of dose-response relationship and kinetic behavior.28 Both of these factors were suggested by Olofsen and Dahan7 as the likely reasons for the difference between their calculated and observed values for t1/2. Recent work confirms that some EEG processing systems, such as BIS A2000 (Aspect Medical Systems, MA) and the M-entropy (GE Health Care) monitors, are not linear and time invariant.4
The studies using processed EEG as a measure of drug effect7,24 were of anesthetized patients. Anesthesia may reduce cerebral bloodflow, and some of the patients studied may have been mildly hypocapnic, which would reduce cerebral perfusion and slow cerebral washin. We monitored end-tidal CO2 in our subjects, and these values remained stable throughout the study period. The kinetics of neuronal dynamics may be different during anesthesia,29 and different central processes have bistable or continuous responses to anesthesia.30 These delaying factors are summarized in Figure 5.
We conclude that the tapping test, which is a simple test of CNS function, could be used to track the onset of effect of nitrous oxide in conscious subjects and allowed us to estimate the time course of cerebral equilibration with a half-time of 2 minutes, which is consistent with other more invasive measures. The DSST is a more sensitive and consistent measure of nitrous oxide action. We plan a further study using an abbreviated form of this test to confirm our present findings with a measure that could be a more appropriate estimate of higher cerebral function.
Name: Gordon B. Drummond, MB, ChB, MA, FRCA, MD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Gordon B. Drummond has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Lauren Bleach, BSc.
Contribution: This author helped design the study, conduct the study, and analyze the data.
Attestation: Lauren Bleach has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Alastair J Thomson, MB, ChB, FRCA.
Contribution: This author helped conduct the study and write the manuscript.
Attestation: Alastair J Thomson has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: R. Ross Kennedy, MB, ChB, PhD, FANZCA.
Contribution: This author helped analyze the data and write the manuscript.
Attestation: R. Ross Kennedy has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
This manuscript was handled by: Tony Gin, FANZCA, FRCA, MD.
a PKPD tools, available at http://www.pkpdtools.com/doku.php/start Accessed December 29, 2011.
1. Eger EI II. Minimum alveolar anesthetic concentration: a standard of anesthetic potency. Anesthesiology 1965; 26: 756–63
2. Ludbrook GL, Upton RN, Grant C, Gray EC. Brain and blood concentrations of propofol after rapid intravenous injection in sheep, and their relationships to cerebral effects. Anaesth Int Care 1996; 24: 445–52
3. Zanner R, Pilge S, Kochs EF, Kreuzer M, Schneider G. Time delay of electroencephalogram index calculation: analysis of cerebral state, bispectral, and Narcotrend indices using perioperatively recorded electroencephalographic signals. Br J Anaesth 2009; 103: 394–9
4. Bibian S, Dumont GA, Zikov T. Dynamic behavior of BIS, M-entropy and neuroSENSE brain function monitors. J Clin Monit Comput 2011; 25: 81–7
5. Duarte R, McNeill A, Drummond G, Tiplady B. Comparison of the sedative, cognitive, and analgesic effects of nitrous oxide, sevoflurane, and ethanol. Br J Anaesth 2008; 100: 203–10
6. Fagan D, Paul DL, Tiplady B, Scott DB. A dose-response study of the effects of inhaled nitrous-oxide on psychological performance and mood. Psychopharmacology 1994; 116: 333–8
7. Olofsen E, Dahan A. The dynamic relationship between end-tidal sevoflurane and isoflurane concentrations and bispectral index and spectral edge frequency of the electroencephalogram. Anesthesiology 1999; 90: 1345–53
8. Turner MJ, McCulloch TJ, Kennedy RR, Baker AB. Pharmacokinetics of sevoflurane uptake into the brain. Anaesthesia 2004; 59: 1201–6
9. Kety SS, Harmel MH, Broomell HT, Rhode CB. The solubility of nitrous oxide in blood and brain. J Biol Chem 1948; 174: 487–96
10. Sase S, Honda M, Machida K, Seiki Y. Comparison of cerebral blood flow between perfusion computed tomography and xenon-enhanced computed tomography for normal subjects: territorial analysis. J Comput Assist Tomogr 2005; 29: 270–7
11. Mcmenemin IM, Parbrook GD. Comparison of the effects of subanaesthetic concentrations of isoflurane or nitrous oxide in volunteers. Br J Anaesth 1988; 60: 56–63
12. Smith G, Shirley AW. A review of the effects of trace concentrations of anaesthetics on performance. Br J Anaesth 1978; 50: 701–12
13. Rampil IJ, Mason P, Singh H. Anesthetic potency (MAC) is independent of forebrain structures in the rat. Anesthesiology 1993; 78: 707–12
14. King BS, Rampil IJ. Anesthetic depression of spinal motor neurons may contribute to lack of movement in response to noxious stimuli. Anesthesiology 1994; 81: 1484–92
15. Yang J, Chai YF, Gong CY, Li GH, Luo N, Luo NF, Liu J. Further proof that the spinal cord, and not the brain, mediates the immobility produced by inhaled anesthetics. Anesthesiology 2009; 110: 591–5
16. Barter LS, Mark LO, Antognini JF. Proprioceptive function is more sensitive than motor function to desflurane anesthesia. Anesth Analg 2009; 108: 867–72
17. Rampil IJ, Kim JS, Lenhardt R, Negishi C, Sessler DI. Bispectral EEG index during nitrous oxide administration. Anesthesiology 1998; 89: 671–7
18. Barr G, Jakobsson JG, Owall A, Anderson RE. Nitrous oxide does not alter bispectral index: study with nitrous oxide as sole agent and as an adjunct to i. v. anaesthesia. Br J Anaesth 1999; 82: 827–30
19. Liley DT, Leslie K, Sinclair NC, Feckie M. Dissociating the effects of nitrous oxide on brain electrical activity using fixed order time series modeling. Comput Biol Med 2008; 38: 1121–30
20. Allison RH, Shirley AW, Smith G. Threshold concentration of nitrous oxide affecting psychomotor performance. Br J Anaesth 1979; 51: 177–80
21. Koblin DD, Dong DE, Deady JE, Eger EI II. Selectiove breeding alters murine resistance to nitrous oxide without alteration in synaptic membrane lipid composition. Anesthesiology 1980; 52: 401–7
22. Ezri T, Sessler D, Weisenberg M, Muzikant G, Protianov M, Mascha E, Evron S. Association of ethnicity with the minimum alveolar concentration of sevoflurane. Anesthesiology 2007; 107: 9–14
23. Doolette DJ, Upton RN, Grant C. Perfusion-diffusion compartmental models describe cerebral helium kinetics at high and low cerebral blood flows in sheep. J Physiol-London 2005; 563: 529–39
24. Rehberg B, Bouillon T, Gruenewald M, Schneider J, Baars J, Urban BW, Kox WJ. Comparison of the concentration-dependent effect of sevoflurane on the spinal H-reflex and the EEG in humans. Acta Anaesthesiol Scand 2004; 48: 569–76
25. Masui K, Kira M, Kazama T, Hagihira S, Mortier EP, Struys MM. Early phase pharmacokinetics but not pharmacodynamics are influenced by propofol infusion rate. Anesthesiology 2009; 111: 805–17
26. Masui K, Upton RN, Doufas AG, Coetzee JF, Kazama T, Mortier EP, Struys MM. The performance of compartmental and physiologically based recirculatory pharmacokinetic models for propofol: a comparison using bolus, continuous, and target-controlled infusion data. Anesth Analg 2010; 111: 368–79
27. Pilge S, Zanner P, Schneider G, Blum J, Kreuzer M, Kochs EF. Time delay of index calculation: analysis of cerebral state, bispectral, and narcotrend indices. Anesthesiology 2006; 104: 488–94
28. Li X, Cui S, Voss LJ. Using permutation entropy to measure the electroencephalographic effects of sevoflurane. Anesthesiology 2008; 109: 448–56
29. Antowiak B. In vitro networks: cortical mechanisms of anaesthetic action. Br J Anaesth 2002; 89: 102–11
30. Lee U, Muller M, Noh GJ, Choi B, Mashour GA. Dissociable network properties of anesthetic state transitions. Anesthesiology 2011; 114: 872–81