Components of general anaesthesia are amnesia, hypnosis, analgesia, muscle relaxation and areflexia . These complex components are usually estimated in real time by observing/measuring simpler parameters such as loss of consciousness (LOC) for hypnosis, movement for analgesia or haemodynamic values for areflexia, upon application of nociceptive stimuli of different intensities and durations.
One of the most difficult issues in everyday anaesthesia practice is its titration for an individual patient and an anticipated/observed level of nociceptive stimulation in order to avoid over- and underdosage. It is common practice to choose an initial dose/concentration of an anaesthetic drug and increase it subsequently if the predefined end-point of anaesthesia (such as LOC) is not reached. Knowledge of the doses/concentrations of anaesthetic drugs associated with defined levels of anaesthesia is relevant for groups of patients but can be difficult to apply to an individual patient. For instance, the predicted propofol concentration to obtain LOC (defined as loss of response to verbal command or LRVC) in 95% of patients (n = 20 patients) has been reported to be 4.16 μg mL−1 with 95% confidence interval (CI) of 3.57–4.74 μg mL−1 . For fentanyl (in the presence of 70% N2O), the Cp50, i.e. the steady-state concentration of the drug associated with lack of movement to skin incision in 50% of patients, was 3.2 ng mL−1 with 67% fiducial limits (±1 SD) of 2.4–4.1 ng mL−1 and an estimated range (n = 18 patients) of 1.5–30 ng mL−1 . These examples illustrate that even in small groups of patients included in clinical studies, pharmacodynamic inter-patient variability (IPV), estimated as the dose or concentration required to reach a predefined end-point of anaesthesia, is high.
In addition to pharmacodynamic IPV, a pharmacokinetic IPV has also been reported. Time to equilibrium between plasma and brain has been calculated from electroencephalography (EEG) derived parameters for several anaesthetic drugs. This allows computer-controlled infusion devices to calculate an average time to equilibrium. IPV for time to equilibrium has been reported. For instance, Scott and colleagues  reported mean T1/2Keo of 6.2 min (range 3.5–11.4 min) for sufentanil and 6.6 min (range 4.7– 10.2 min) for fentanyl. These examples illustrate that there are both pharmacodynamic and pharmacokinetic sources of IPV.
How can this information, derived from clinical pharmacology studies, be applied to an individual patient? For instance, when providing anaesthesia for an elderly patient, even with sophisticated tools such as target-controlled infusion (TCI) devices, what initial target should be chosen by the anaesthesiologist in order to reach a predefined end-point such as LOC? One strategy proposed in clinical trials is the ‘up-and-down’ technique  but this cannot be applied to an individual patient in everyday clinical practice. Another option would be to choose the EC 95 value that is reported to be associated with LOC. This would result in a 5% probability that the value is not high enough but a much higher probability that the patient is overdosed and at risk of cardiovascular instability. Another approach would be to choose a lower dose/ target concentration and progressively increase the dose/target until the adequate value is found. This approach would lower the risk of cardiovascular instability but would increase the duration of anaesthesia induction.
Whatever the approach used, they are all based on the assumption that for a given choice (weight-adjusted dose or TCI-based calculated concentration), the time to reach the predefined end-point are similar within a group of patients. In order to test this hypothesis and attempt to improve anaesthetic drugs titration algorithms, we designed the present study in which patients were given a similar anaesthetic regimen and the time to reach several end-points of anaesthetic depth were measured. One end-point of interest, in addition to LOC and haemodynamic parameters such as arterial pressure and heart rate (HR), is a processed EEG-derived parameter obtained with the BIS® monitor (Aspect Medical System Inc., Newton, MA, USA) . Bispectral index scale (BIS) values decrease linearly with increasing concentrations of hypnotics and provide additional information not available from the clinical end-points mentioned above .
Materials and methods
Induction with sevoflurane and the SiBI® connector  are routinely performed in adult patients in our institution. This induction technique is not offered to patients with significant obesity (body mass index or BMI > 35), history of problems with inhaled anaesthetics, asking for total intravenous (i.v.) anaesthesia, at risk of aspiration or malignant hyperthermia. Patients are informed of the details of this induction technique at the preoperative visit.
Thirty patients, ASA I–II, aged 18–60 yr, without cardiac, pulmonary, hepatic, renal or neurological impairment were considered eligible for this observational study after informed consent. Their ‘selection’ was based on the presence in the operation room of the first author (PL). There were no changes in the anaesthetic regimen because of the study. The patients were scheduled for cervicotomy, sacral or axillary subcutaneous surgery, inguinal hernia repair or bone marrow sampling.
Premedication was not standardized and performed with hydroxyzine or alprazolam 12 and 1 h before induction of anaesthesia, except for three patients taking chronically other sedative drugs. Three patients were not premedicated.
After arrival in the operating room, a peripheral i.v. cannula was inserted in an antecubital vein and 10 mL kg−1 of Ringer's lactate solution was infused before onset of induction. The montage of the i.v. line is standardized in our institution when remifentanil is used. It consists of a connector linked to the i.v. cannula (dead volume 4.2 mL, Braun, Melsunger, Germany) followed by a one-way valve (Vygon, Ecouen, France) linked to the remifentanil-filled syringe by a 2 m connector (Vygon, Ecouen, France) with a dead volume of 2 mL.
Standard monitoring (electrocardiogram, non-invasive blood pressure, arterial oxygen saturation), was performed with Spacelab® monitors (Spacelab Medical, Redmond, WA, USA). Recorded HR and mean arterial pressure (MAP) values were those displayed by the monitor. After skin preparation, a BIS® XP sensor (Aspect Medical Systems, Natick, MA, USA) was applied to the forehead skin according to the manufacturer's instruction. The sensors were placed to explore either the left or the right frontal cortex; EEG was recorded and analysed using an A2000 XP BIS® monitor (Aspect Medical Systems, Natick, MA, USA). Respiratory gases were sampled at the SiBI® connector, and inspired and end-tidal (ET) oxygen, carbon dioxide and sevoflurane (after onset of induction) concentrations were continuously monitored.
Preoxygenation via a good-fitting face mask was performed for 4 min in pure oxygen using the SiBI® connector and an additional source of oxygen . During preoxygenation, the circuit of a SA2® (Drager, Lübeck, Germany), a Cato® (Drager, Lübeck, Germany) or a Julian® (Drager, Lübeck, Germany) anaesthesia workstation was primed with 6% sevoflurane and oxygen with a fresh gas flow 6 L min−1. Analysis of inspired and expired gases was performed with the gas analyser PM8050® (Drager, Lübeck, Germany; associated with the SA2® ventilator) and the integrated gas analysers of the Cato® and Julian® anaesthesia machines. All gas analysers were regularly calibrated according to the manufacturers' instructions.
Baseline values for BIS, MAP and HR were those recorded at the end of the preoxygenation period. Subsequently, values were recorded manually every minute for a period that finished at least 3 min after tracheal intubation or until the nadir values of MAP or HR were reached, whichever occurred first. Onset of induction was marked by turning the flow selector of the SiBI® connector to the induction position. LOC was defined as LRVC which was evaluated every 15 s following onset of induction.
When the BIS value reached 60, the delivered concentration of sevoflurane was decreased to 4% and remifentanil (diluted at a concentration of 20 μg mL−1) was infused using an Alaris® (Alaris Medical Systems, Hampshire, Great Britain) syringe pump starting with a 0.5 μg kg−1 min−1 bolus injected over 60 s followed by a continuous infusion of 0.25 μg kg−1 min−1 (Fig. 1).
In case of non-obstructive apnoea, lungs were ventilated manually. Laryngosopy and tracheal intubation were performed 4 min after starting the remifentanil infusion. After intubation, the sevoflurane delivered concentration was decreased to 2% in a fresh gas flow of 2 L min−1 containing 50% oxygen with air; tidal volume was 8 mL kg−1 and respiratory rate 15 breaths min−1 (Fig. 1).
If HR decreased to values below 50 beats min−1, 10 μg kg−1 of atropine was injected; if MAP decreased to values below 50 mmHg, ephedrine 6 mg was injected.
Computer simulations of the delivered, inspired, ET and predicted brain concentrations of sevoflurane were performed using the GasMan® software (Med Man Simulations Chesnut Hill, MA, USA, written by Dr J. H. Philip, Harvard Medical School, Boston, USA; http://www.gasmanweb.com). Saturation of the anaesthesia circuit with sevoflurane was simulated. The fresh gas flow was at 6 L min−1. The body weight and age were fixed at the average values observed in the patients included in the study (see below). The mean, minimum and maximum values for the latencies to reach defined end-points of depth of anaesthesia were superimposed on the graph of the simulation.
Statistical analysis was performed with the SigmaStat® for Windows version 3.0 (version 3.0 for Windows, SPSS Inc., USA). The results are presented as mean ± SD or as median (minimum–maximum). The latencies to reach the defined clinical, BIS and haemodynamic end-points were compared using analysis of variance (ANOVA) followed by the PLSD test for multiple paired comparisons. When the values did not follow a normal distribution, the latencies were compared with non-parametric analysis of variance followed by the Dunnett's test for multiple paired comparisons.
Seventeen males and thirteen females were included in this observational study after approval by the local Ethics Committee. Age, weight and BMI by gender for all the thirty patients are presented in Table 1.
Values of BIS spread widely at different end-points (Table 2). At LRVC, BIS values were 86 ± 7 varying from 71 to 98. The nadir values of BIS after induction were 29 ± 8 varying from 17 to 47. The mean value of BIS before laryngoscopy was 39 (±14) varying from 17 to 69. Within 2 min after intubation the mean value of BIS was 46 (±15) varying from 23 to 80.
Latencies to reach the different end-points varied among individuals (Table 3). Ratio (highest to lowest values) for the latencies to reach LRVC, a BIS value of 60, a BIS value of 40 and the nadir BIS value varied from 3 to 6 in this group of 30 patients. The ratios increased with deeper levels of hypnosis.
Mean ET sevoflurane (ETSevo) concentrations for a BIS value of 90, LRVC, a BIS value of 60, a BIS value of 40, the minimal BIS value and before laryngoscopy are presented in Table 4.
We investigated possible factors that could predict the time necessary to reach a BIS value of 60 and calculated correlation coefficients between several independent (predictive) parameters and the measured time to reach a BIS value of 60 as the dependent (predicted) parameter (Table 5). The only predictive parameter that was statistically correlated (R2 = 0.41; P < 0.05) with the time to reach a BIS value of 60 was the time to reach LRVC (Fig. 2). Nevertheless, the 95% CI for the slope of the regression line is (0.9–2.4). The latencies to reach a BIS value of 60 were not correlated to: (i) parameters such as age, weight, BMI; (ii) the ETSevo values measured when BIS value was 60 (as expected because the alveolar and brain concentrations were not in equilibrium); (iii) baseline HR value as a surrogate of cardiac output . There was no effect of the premedication or of the drug used for premedication on the time to reach LRVC (P = 0.9; ANOVA) or a BIS value of 60 (P = 0.8; ANOVA on ranks).
Time to reach the nadir value of MAP ranged from 120 to 1200 s and from 300 to 1020 s for HR (Table 3). One patient had a nadir value for MAP between 40 and 49 mmHg. Thirteen patients had nadir values for MAP between 50 and 59 mmHg. Twenty-four patients had nadir values of MAP between 60 and 69 mmHg. One patient required a bolus of ephedrine; six patients had nadir values of MAP above 70 mmHg. Fourteen patients required a bolus infusion of atropine because of bradycardia.
The latencies to the nadir values of BIS value were significantly (P < 0.05) shorter than the latencies required to reach the nadir values of MAP (Table 3). The latencies to the nadir values of BIS and of MAP were correlated (R2 = 0.25; P = 0.004); latencies to the nadir values of MAP and of HR were strongly correlated (R2 = 0.62; P < 0.001).
Minimum values of BIS and MAP were not correlated (R2 = 0.05; P > 0.05). Time to the minimal values of MAP and the percentage of MAP decrease (nadir vs. baseline) were independent of age.
The simulation of the inspired, ET and vascular rich organs (to which the brain is assimilated) concentrations using the GasMan® software is presented in Figure 3. The median and extreme values for the time to reach predefined end-points, as well as the measured ETSevo concentrations were superimposed on the simulation graph.
The aim of the present study was to test the hypothesis that a similar anaesthetic regimen would result in similar latencies to reach predefined end-points of anaesthetic depth. The main finding of this study is that this assumption is not valid. IPV, estimated by the latency required to reach predefined end-points of depth of anaesthesia, is clinically important. The ratios of the longest to the shortest time for 30 patients are of 3 for LRVC and increased to 6 at the nadir of BIS values. Other important findings of this study are that: (i) the latencies to LRVC and a BIS value of 60 were correlated (R2 = 0.41); (ii) the latencies to reach a BIS value of 60 were independent of age, BMI, weight, and HR measured before induction (as an estimate of cardiac output ), premedication regimen and also independent (as expected because equilibrium between arterial and effect site concentrations were not reached) from the ETSevo measured when the BIS value was 60; (iii) the latencies to reach the nadir values of BIS value were significantly shorter than the latencies to the nadir values of MAP; (iv) the nadir values of BIS and the nadir values of MAP were not correlated and were independent of age.
What are the clinical implications of our results?
Titration of anaesthesia for an individual patient and a defined level of nociceptive stimulation is still challenging and there are not many published titration algorithms. Titration of anaesthesia upon induction is highly desirable to avoid both cardiovascular instability due to anaesthetic drug overdose but also explicit awareness due to too low effect site (brain) concentrations of hypnotics. Several recent reports suggest that short durations (dozens of seconds or a few minutes) of inadequately low concentrations of anaesthetic drugs (in the presence or absence of a nociceptive stimulus [9,10]) are sufficient to result in explicit awareness upon induction and tracheal intubation [10,11]. If one wants to avoid both over- and underdosage of anaesthetic drugs, an important question is when should one apply a nociceptive stimulus such as laryngoscopy and tracheal intubation upon induction? Answers to this question are few in the literature. For instance, in a study with a design similar to the one we used, 20 adults patients (19–32-yr old) were randomized to receive 6–7 vol.% sevoflurane with 6 L min−1 fresh gas flow with or without 66% nitrous oxide/28% oxygen; three vital capacity breaths were performed thereafter ventilation was manually assisted . The time of exposure to the inhaled gas was varied for consecutive participants using the ‘up-and-down’ method of Dixon . Mean time to reach loss of lid-lash reflex was 1.1 min with a 95% CI of 0.5–1.7 min. Mean time for pupils to converge was 4.8 min with a 95% CI of 2.8–6.8 min. These results suggest that trying to apply a nociceptive stimulus after a fixed interval of time (a procedure we call dose-based titration) in all patients during induction may be an inadequate strategy. Our results demonstrate that this would result in different BIS values even for small groups of patients. Superposition of the median and extreme values of the latencies to reach a BIS value of 60 on the GasMan® simulation clearly shows that at a fixed time point (Fig. 3), many patients do not reach a BIS value of 60 (for which the probability of explicit awareness is low or zero ). Such application of a potent nociceptive stimulus at a moment when effect site concentrations of anaesthetic drugs were too low (and probably the BIS values too high) may explain the high incidence of expiratory stridor reported by Muzi and colleagues . In the patients studied in the present report there were no such complications probably because manipulation of the airway was initiated only when the BIS values reached 60. Clearly, our results suggest that in addition to the classic dose-based titration approach, integrating the wide IVP for the time to reach a predefined end-point of anaesthetic depth (a procedure we would like to call time-based titration) could be helpful in defining algorithms of anaesthetic drugs titration for an individual patient). One helpful end-point for time-based titration is the latency to reach a BIS value of 60.
An important question raised by our observations is whether the time to reach a BIS value of 60, for a given patient, could be predicted by other pieces of information such as age, body weight, the HR as an estimate of cardiac output, the type of premedication or the measured concentrations of inhaled anaesthetics. The answer is clearly no; the only clinically available information that predicted the latency to reach a BIS value of 60 is (as expected because of the BIS monitor algorithm) the latency to LRVC (Fig. 2). Nevertheless, if one integrates the information that a few minutes of inadequately high BIS values are sufficient to result in explicit awareness [9,10], the CI of the regression equation linking the two latencies still suggests that the only reasonable alternative (except from anaesthetic drug overdose) to avoid explicit awareness is to wait until the BIS values reach 60.
Another option to avoid inadequately low brain concentrations of anaesthetic drugs would be to wait until arterial pressure reaches its nadir value. In the present study, nadir values of MAP and HR were reached significantly later than the nadir values for BIS. This has also been reported by Kazama and colleagues during propofol TCI in 41 patients . Our results suggest that if the BIS value is chosen for titration during induction, this can require a significantly shorter interval for titration. Interestingly, there was no correlation between the nadir values of BIS and of MAP values suggesting, as already reported by other investigators , that MAP values cannot replace BIS values for titration of anaesthesia.
A limitation of our study, due to its observational design is that five patients did have BIS values above 60 immediately after tracheal intubation despite infusion of remifentanil. In this context, the increased BIS values can be interpreted as inadequate analgesia . Avoiding such increases of the BIS value would have required a careful titration of the opioid as reported by our group in cardiac surgery patients . In that study, a titration algorithm of anaesthetic drugs based on information provided by TCI devices for both sufentanil and propofol and by the BIS monitor was used. Its goal was to find the lowest concentration of sufentanil that would avoid an increase in BIS value above 60 upon tracheal intubation . Another limitation of this study is that we did not take into account, because of the observational design, factors such as the vascular or endocrine status of the patient. Nevertheless, the key message of our study is that the approach based on time-dependent titration would reflect most of the factors responsible for this variability.
By taking into consideration the previously published algorithm  and the results presented in the present study, we propose that the BIS monitor could be helpful to improve everyday anaesthesia practice not as a predictive tool as suggested by other investigators that clearly showed its limits  but as a tool for time-based titration. In this context, the recently published results of Myles and colleagues  and Monk and colleagues  clearly demonstrate that a strategy of titration of anaesthetic drugs can result in improved quality of care by reducing morbidity (decreased incidence of explicit awareness by avoiding BIS values above 60) and possibly reducing mortality (by avoiding anaesthetic drugs overdose and too low BIS values).
1. Drummond JC. Monitoring depth of anesthesia: with emphasis on the application of the bispectral index and the middle latency auditory evoked response to the prevention of recall. Anesthesiology
2. Iselin-Chaves IA, Flaishon R, Sebel PS et al
. The effect of the interaction of propofol and alfentanil on recall, loss of consciousness, and the Bispectral Index. Anesth Analg
3. Glass PS, Doherty M, Jacobs JR, Goodman D, Smith LR. Plasma concentration of fentanyl, with 70% nitrous oxide, to prevent movement at skin incision [see comments]. Anesthesiology
4. Scott JC, Cooke JE, Stanski DR. Electroencephalographic quantitation of opioid effect: comparative pharmacodynamics of fentanyl and sufentanil. Anesthesiology
5. Dixon WJ. Staircase bioassay: the up-and-down method. Neurosci Biobehav Rev
6. Rampil IJ. A primer for EEG signal processing in anesthesia. Anesthesiology
7. Colas MJ, Tetrault JP, Dumais L, Truong P, Claprood Y, Martin R. The SiBI connector: a new medical device to facilitate preoxygenation and reduce waste anesthetic gases during inhaled induction with sevoflurane. Anesth Analg
8. Mertens MJ, Olofsen E, Burm AG, Bovill JG, Vuyk J. Mixed-effects modeling of the influence of alfentanil on propofol pharmacokinetics. Anesthesiology
9. Ekman A, Brudin L, Sandin R. A comparison of bispectral index and rapidly extracted auditory evoked potentials index responses to noxious stimulation during sevoflurane anesthesia. Anesth Analg
10. Myles PS, Leslie K, McNeil J, Forbes A, Chan MT. Bispectral index monitoring to prevent awareness during anaesthesia: the B-Aware randomised controlled trial. Lancet
11. Ekman A, Lindholm ML, Lennmarken C, Sandin R. Reduction in the incidence of awareness using BIS monitoring. Acta Anaesthesiol Scand
12. Muzi M, Robinson BJ, Ebert TJ, O'Brien TJ. Induction of anesthesia and tracheal intubation with sevoflurane in adults. Anesthesiology
13. Kazama T, Ikeda K, Morita K et al
. Comparison of the effect-site k(eO)s of propofol for blood pressure and EEG bispectral index in elderly and younger patients. Anesthesiology
14. Flaishon R, Windsor A, Sigl J, Sebel PS. Recovery of consciousness after thiopental or propofol. Bispectral index and isolated forearm technique. Anesthesiology
15. Guignard B, Menigaux C, Dupont X, Fletcher D, Chauvin M. The effect of remifentanil
on the bispectral index change and hemodynamic responses after orotracheal intubation. Anesth Analg
16. Forestier F, Hirschi M, Rouget P et al
. Propofol and sufentanil titration with the bispectral index to provide anesthesia for coronary artery surgery. Anesthesiology
17. Struys MM, Jensen EW, Smith W et al
. Performance of the ARX-derived auditory evoked potential index as an indicator of anesthetic depth: a comparison with bispectral index and hemodynamic measures during propofol administration. Anesthesiology
18. Monk TG, Saini V, Weldon BC, Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg