The combination of epidural with general anesthesia is a widely used technique, which is thought to reduce intraoperative general anesthetic requirements. Evidence for such a reduction is limited because of difficulties in reliably determining anesthetic depth.
Although epidural anesthesia reduces the minimum alveolar anesthetic concentration (1), a spinal phenomenon, the implications of this finding for intraoperative consciousness are unclear. Postoperative arousal is delayed by epidural lidocaine in patients receiving equivalent intraoperative isoflurane doses (2). Intraoperative titration of isoflurane administration to anesthetic depth by some means might reasonably be expected to prevent this effect. In a pilot study, we aimed to identify the intraoperative electroencephalographic (EEG) variable that best distinguished patients receiving combined epidural and general anesthesia from those receiving general anesthesia alone. We established that EEG 95% spectral edge frequency (SEF) and median frequency (MF) are decreased in patients with epidural blockade, but bispectral index (BIS) is unaffected (Appendix 1). In our principal study, presented below, we tested the hypothesis that epidural blockade reduces general anesthetic requirements for maintenance of equivalent intraoperative SEF.
Both pilot and principal studies received Ethics Committee approval, and written, informed consent was obtained from all patients. In the pilot study, 30 patients were studied to determine the best target EEG variable for anesthetic administration by closed-loop control (Appendix 1). On the basis of our results, we studied 50 ASA status I–II patients, aged 20–65 yr, who were undergoing abdominal hysterectomy. Patients with contraindications to epidural anesthesia or a hemoglobin concentration <10 g/dL were excluded, as were patients receiving drugs with central nervous system effects. Patients were randomly allocated by selection of sealed envelopes to one of two groups, epidural or control, and no premedication was given.
In epidural patients, epidural anesthesia was established before surgery. Intravenous crystalloid solution 500 mL was administered as a preload. A lumbar epidural catheter was inserted under local anesthesia with 2% lidocaine. An epidural test dose of 4 mL of 0.5% bupivacaine with 1:200,000 adrenaline was administered, followed, in the absence of adverse effects, by 10 mL of epidural 0.5% bupivacaine. Sensory blockade to ice was assessed at 5-min intervals until it extended rostral to the operative site. Patients in the control group received no preoperative treatment.
In all patients, hemodynamic monitoring was commenced before anesthetic induction by using a Narkomed 4E anesthesia machine with an attached Vitalert 2000 monitor (North American Drager, Telford, PA). Silver-silver chloride gel-filled electrodes were applied at Fp1, Fpz, and Fp2, with a ground electrode on the forehead. Bifrontal EEG recording was commenced before the induction by using an Aspect A-1000 EEG monitor (software Version 3.12; Aspect Medical Systems, Framingham, MA). Electrode impedance was maintained at <5 kΩ. Heart rate (HR) and end-tidal carbon dioxide (P E′CO2) and isoflurane concentrations (F E′ISO) were downloaded from the RS-232 port of the anesthetic machine at 5-s intervals, and noninvasive mean arterial blood pressure (MAP) was downloaded at 5-min intervals. SEF, MF, and BIS were downloaded at 5-s intervals from the “Processed EEG” port on the Aspect monitor.
The software application Monitor was used, designed for a Macintosh PowerPC (Apple Computer, Inc., Cupertino, CA) computer by one of the authors (JD). Monitor enables the user to download data from several sources into the same file by using a multiple serial interface (SEQS; Creative Solutions Inc.). The application incorporated a modified proportional-integral-derivative controller algorithm for drug administration to allow closed-loop control of isoflurane administration with a target EEG value. We have used a similar technique for closed-loop administration of both IV and volatile anesthesia, targeting BIS (3). This allowed the computer to control anesthetic drug administration by using a syringe pump (Graseby 3400; Graseby Medical, Watford, England). Before commencing the study, the system was tuned to an SEF of 17.5 Hz by using the method of Ziegler and Nichols (4). We considered this to be an appropriate intraoperative target on the basis of the pilot study, in which the administration of 1%F E′ISO in the control group resulted in a similar SEF.
The study protocol, and the algorithm for isoflurane administration with which the closed-loop controller was programmed, were adjusted after the 17th patient. After surgery, this control group patient reported momentary awareness at incision. Data collected from the patient were analyzed immediately. It appeared that SEF before the induction was unusually low (10 Hz), despite a BIS of 90, and stayed below the set point of 17.5 Hz throughout the induction until just before intubation. Thereafter it remained at approximately 19.5 Hz until incision. After the initial bolus of isoflurane, given shortly after IV induction, the controller administered isoflurane at the minimum rate despite an increasing SEF, because of the long preceding period during which the SEF was <17.5 Hz. This phenomenon, well described in engineering literature (5), is known as integral windup. When the investigator (APM) observed a marked tachycardia after incision in this patient, a bolus dose of isoflurane was given by using a manual override function built into the computer software. For subsequent patients, the controller was programmed to ignore SEF data before the detection of isoflurane in the anesthetic circuit, to minimize the risk of integral windup.
Anesthesia was induced with propofol 1.5 mg/kg. IV alfentanil was infused from the induction onward to attain a target plasma concentration of 100 ng/mL. IV vecuronium 0.1 mg/kg was given initially, followed by an infusion, adjusted to maintain neuromuscular blockade at one twitch in a train-of-four sequence. Patients were intubated and ventilated with 30% oxygen in air at a fresh gas flow of 3 L/min via a circle absorber system.
In the first 17 patients, the closed-loop controller was switched on immediately after intubation, and an initial dose of 0.015 mL/kg of isoflurane was injected into the inspiratory limb of the circle absorber system from the syringe pump. In the remaining 33 patients, investigators delayed starting the controller until the SEF began to increase after IV induction. Thereafter, the intraoperative isoflurane infusion rate was dictated by the controller to maintain SEF at 17.5 Hz. At intervals of approximately 20 min, venous blood samples were taken, and assays of plasma bupivacaine and alfentanil concentrations were subsequently conducted by use of standard methods.
During surgery, all patients lay on a warming mattress. Body temperature data were collected in 40 patients with a nasopharyngeal temperature probe, which was inserted after the induction. Warmed IV normal saline was administered during surgery to replace the fluid deficit from preoperative fasting, continuing insensitive losses, and blood loss from surgery. When MAP decreased to less than 75% of baseline, 1 mg of IV metaraminol, a noncatecholamine sympathomimetic with small potential for crossing the blood-brain barrier, was given. When HR decreased to less than 50 bpm, 0.2 mg of glycopyrrolate was given IV.
Patient management was standardized after stopping isoflurane administration on closure. Ventilation was continued with 100% oxygen at a fresh gas flow of 6 L/min, and patients were not physically stimulated. Every 30 s, they were instructed at normal speaking volume by an observer blinded to group allocation to open their eyes. If adequate spontaneous ventilation was established, they were then tracheally extubated. The time from stopping isoflurane administration to eye opening was recorded as a clinical indicator of the preceding intraoperative anesthetic depth. In the postanesthesia care unit, 15 min after eye opening, patients were asked to describe their pain on a four-point scale: none, 0; mild, 1; moderate, 2; or severe, 3. They were asked about dreaming and awareness and were interviewed again at 24 h after surgery.
Data were analyzed with Stata statistical software (Release 6.0; Stata Corp., College Station, TX). Patient characteristics and operative details were compared between groups by using two-sample Student’s t-tests and χ2 tests. A mean intraoperative plasma alfentanil and bupivacaine concentration was calculated for each patient.
In each patient, a mean intraoperative value for F E′ISO, body temperature, P E′CO2, MAP, HR, and each EEG variable was calculated from the 5-s interval data between the induction and the point at which isoflurane administration was stopped. Group means for these intraoperative variables were compared by using two-sample Student’s t-tests with unequal variance. Binomial data were compared by using Fisher’s exact test and risk ratios.
To determine which variables might be good predictors of time to eye opening, a mean value for each EEG variable (SEFfinal, MFfinal, and BISfinal) and for F E′ISO (F E′ISOfinal) was calculated from 5-s interval data for the final 30 s of isoflurane administration. These values were plotted against time to eye opening, and linear regression was performed for each of the four variables. Correlation coefficients were then compared by using the Stata command “corcor”(6).
Postoperative pain scores were compared by us-ing a Mann-Whitney U-test. Associations between intraoperative F E′ISO and intraoperative MAP, metaraminol dosage, body temperature, dreaming, and postoperative pain were subjected to multiple regression analyses.
Demographic data and operative details are presented in Table 1. Similar numbers of patients in each group underwent surgical procedures in addition to hysterectomy. These were mostly bilateral salpingo-oophorectomy, but omentectomy was performed on two patients in the epidural group and appendectomy on one control patient. The mean intraoperative SEF did not differ significantly between groups and was between 17 and 18 Hz in all but five patients. In three patients (one epidural and two controls), the mean SEF was below this range, and in two (one epidural and one control) it was above.
Intraoperative data are presented in Table 2. Continuous data are normally distributed. The mean (sd) intraoperative plasma bupivacaine concentration in epidural patients was 528 (217) ng/mL. Between-group differences were not significant for plasma alfentanil concentrations, P E′CO2, HR, or MF. To maintain an intraoperative SEF of approximately 17.5 Hz in both groups, a smaller F E′ISO was required in epidural patients. Intraoperative body temperature and MAP were also decreased in the epidural group. Adjusting for the effect of epidural anesthesia on F E′ISO, a decrease in body temperature of 1°C resulted in a change of −0.03% in F E′ISO (95% confidence interval [CI], −0.20% to 0.14%;P = 0.72), and a decrease in MAP of 10 mm Hg resulted in an increase of 0.01% in F E′ISO (95% CI, −0.05% to 0.06%;P = 0.78). This suggests that temperature and MAP do not account for between-group differences in F E′ISO. Similarly, F E′ISO did not appear to be related to metaraminol dosage. BIS was higher in control than in epidural patients.
Data relating to eye opening and the postoperative interview are shown in Table 3. Some data at eye opening were lost in one patient from the epidural group, who woke 30 s after isoflurane administration was stopped. Relationships between time to eye opening and SEFfinal, MFfinal, BISfinal, and F E′ISOfinal are illustrated in Figure 1. Of the four variables, only F E′ISOfinal and BISfinal correlated with time to eye opening, with the correlation of F E′ISOfinal being significantly the better of the two (P < 0.0001).
Ten patients in the control group reported moderate or severe pain. Epidural patients were either pain free or reported only mild pain. Adjusting for the effect of group on F E′ISO, a one-point increase in postoperative pain score was associated with a mean reduction of 0.04% in intraoperative F E′ISO (95% CI, −0.12% to 0.03%;P = 0.26). This suggests that there is no relationship between high postoperative pain scores and large intraoperative isoflurane requirements. The mean F E′ISO in patients reporting dreams was 0.1% smaller than in patients who did not (95% CI, −0.28% to 0.07%;P = 0.24). One patient in the control group reported awareness at the point of incision, as described previously. The patient was not apparently distressed on subsequent interviews over the next few days and reported no postoperative symptoms related to her experience.
In this study, we found that epidural blockade reduces intraoperative isoflurane requirements by 21% when SEF 17.5 Hz is used as a target for closed-loop isoflurane administration. Delayed postoperative arousal as result of epidural blockade, observed by other authors in similar clinical circumstances (2), was thereby avoided. The reduced isoflurane requirement in our epidural patients is explained by the fact that SEF is suppressed more by combined epidural and general anesthesia than by general anesthesia alone, as seen in our pilot study.
There are several possible mechanisms, direct and indirect, by which epidural blockade might have affected intraoperative SEF, and thence isoflurane administration, via the closed-loop controller. The first of the potential direct mechanisms, and the most intuitive, is attenuation of nociceptive input. Were this the explanation, we might have expected high postoperative pain scores in our patients to be associated with large intraoperative isoflurane requirements, a relationship that we were unable to demonstrate. Second, systemically absorbed bupivacaine in the epidural group may have been responsible. However, in patients recovering from isoflurane anesthesia, delayed postoperative arousal and reduced MAC-Awake of isoflurane are seen only in those receiving epidural, and not IV, lidocaine (2). Third, local anesthetics injected into the lumbar epidural space might have diffused cranially in the cerebrospinal fluid and exerted a direct effect on the cerebral cortex. Finally, and most likely in our opinion, epidural effects may be independent of nociceptive stimuli and may be related instead to sensory deafferentation. In a previous study, we showed that epidural blockade alone, in the absence of nociceptive stimuli, produces no significant EEG changes but that the sensory deafferentation produced by spinal blockade, arguably denser, does (7). Spinal blockade also reduces thiopentone, midazolam (8), and propofol (9) hypnotic requirements.
The effect of epidural blockade on SEF, and thence F E′ISO, might have been mediated indirectly. Hemodynamic effects of combining epidural with general anesthesia, as observed in our study and others (10,11), may be relevant. Changes in cerebral blood flow, as a result of decreased MAP in epidural patients, may have affected SEF, although previous authors have failed to demonstrate such an effect (12,13). Having adjusted for the effects of epidural anesthesia, we found no evidence that the difference in F E′ISO could be explained by differences in intraoperative metaraminol dosage or body temperature.
To conclude that the 21% reduction in isoflurane requirements for SEF suppression is equivalent to the hypnotic effect of epidural anesthesia relies on the assumption that SEF accurately measures the hypnotic state. Such a conclusion would be wrong without demonstrating that the clinical hypnotic effect was the same in the two groups. Two potential clinical indicators of this effect were the incidence of dreaming and the time from stopping anesthetic administration to eye opening. There was no difference in the incidence of dreaming between groups. Time to eye opening was shorter in the epidural group, consistent with smaller F E′ISO, suggesting that these patients may have been more lightly anesthetized than controls. If this was the case, despite closed-loop maintenance of the same SEF in each group, it implies that the relationship between SEF and clinical hypnosis during isoflurane anesthesia may be affected by epidural blockade.
Should we have used BIS as our closed-loop target instead? Studies on spontaneously breathing subjects have demonstrated the predictive superiority of BIS over other EEG-derived variables, including SEF, for several clinical end-points and anesthetics (14–18). The implications of these findings for intraoperative patients with neuromuscular blockade are unclear at present, because ethical considerations have made relevant investigations difficult. In our pilot study, intraoperative BIS did not differ between groups, and BIS during desflurane anesthesia has also been shown to be unaffected by epidural bupivacaine 0.125%(19). This raises the interesting possibility that, had we used BIS as a closed-loop target in our principal study, there might have been no between-group difference in isoflurane requirements.
Intraoperative BIS was higher in epidural patients, reflecting better than SEF the lighter hypnotic state caused by smaller F E′ISO. However, the variable in the final 30-second period of anesthesia that correlated better with time to eye opening than any of the EEG variables, including BIS, was F E′ISO. This finding accords with that of Glass et al. (20), who did not find BIS superior to F E′ISO in predicting sedation score during isoflurane administration.
Some aspects of our experimental method merit further discussion. Most patients in our pilot study were male, whereas all in the principal study were female. This difference might be relevant. Evidence published after data collection had been completed indicates that women emerge faster from general anesthesia than men (21). However, multiple regression analysis of pilot study data does not support a sex effect on SEF.
In our principal study, investigators were not blinded. However, potential investigator bias was minimized because data recorded by monitors were downloaded directly onto a laptop computer, and criteria for treatment in the event of intraoperative hypotension or bradycardia, and treatment itself, were predetermined. The use of a closed-loop controller meant that investigators had no influence over intraoperative isoflurane administration.
Regarding closed-loop techniques involving the EEG, these have been used successfully for anesthetic administration with both MF (22,23) and BIS (4,24) as targets. The occurrence of an episode of awareness in our study demonstrates some limitations of closed-loop control. Increased dosage can occur only in response to, rather than in advance of, episodes of surgical stimulation or light anesthesia. The appropriateness and timing of such a response depends on the reliability of the target variable and the algorithm with which the controller is programmed. In our aware patient, the preinduction SEF was 10 Hz, a figure more usually associated with deep anesthesia and interpreted as such by the controller. We were able to modify the study protocol and the controller software so as to continue the study with little risk of a similar episode. However, we cannot recommend SEF as a target for closed-loop control of anesthesia.
In conclusion, patients receiving intraoperative combined epidural and general anesthesia require 21% less isoflurane to maintain an SEF of 17.5 Hz than those receiving general anesthesia alone, and they consequently wake faster. If epidural anesthesia contributes to the hypnotic effect when a combined technique is used, it must presumably be to a lesser degree than 21%. Even with smaller F E′ISO, patients receiving combined regional and general anesthesia have a lower MAP and are more likely to require treatment for hypotension than those receiving general anesthesia alone. Anesthesiologists who use a combined technique similar to ours will frequently encounter hypotension. The decision to reduce isoflurane dosage in these circumstances should be made with the knowledge that epidural anesthesia confers little additional hypnotic effect. Substantial reductions in isoflurane dosage in these patients may lead to an increased risk of awareness.
In a pilot study, 30 ASA I patients, aged 20–65 yr, who were undergoing major lower-limb orthopedic surgery, were randomly allocated to epidural or control groups. In epidural patients, epidural blockade was established as described previously. In all patients, general anesthesia was induced and maintained with isoflurane in oxygen/air, by using a manually adjusted vaporizer, and alfentanil and vecuronium were administered as described previously. Data acquisition and general intraoperative management were as described previously. Fifteen minutes were allowed to elapse at F E′ISO 0.75% between intubation and incision, and a further 15 min at F E′ISO 0.75% after incision. Depending on the operative duration, patients then received a further 2 to 4 randomly ordered F E′ISO values (0.5%, 1%, 1.25%, or 1.5%), each for 30 min. F E′ISO was increased by 0.5% if patients displayed signs of inadequate anesthesia. After data collection was complete, patients were managed as for a standard anesthetic. For each variable analyzed, the first 15 min of data recorded at each F E′ISO were discarded, to account for equilibration, and an average value for each patient was calculated from the remainder. Multiple regression analysis using general estimating equations was used to compare the two groups. Age, weight, height, sex distribution (M/F, 11:4 [controls] and 14:1 [epidurals]), and duration of anesthesia were similar. Intraoperative between-group differences appear in Table 4. Of the two EEG variables significantly affected by epidural blockade, we chose SEF as a target variable for the principal study, because MF exhibited wide interindividual variability at small F E′ISO; i.e., at the intended F E′ISO 0.5%, mean (sd) SEF and MF values for all patients were 21.2 (2.7) Hz and 10.5 (3.7) Hz, respectively.
1. Hodgson PS, Liu SS, Gras TW. Does epidural anesthesia have general anesthetic effects? Anesthesiology 1999; 91: 1687–92.
2. Inagaki Y, Mashimo T, Kuzukawa A, et al. Epidural lidocaine delays arousal from isoflurane anesthesia. Anesth Analg 1994; 79: 368–72.
3. Morley A, Derrick J, Mainland P, et al. Closed loop control of anaesthesia: an assessment of the bispectral index as the target of control. Anaesthesia 2000; 55: 953–60.
4. Ellis G. Control system design guide. New York: Academic Press, 1991: 76–111.
5. Åström KJ, Hägglund T. PID controllers: theory, design, and tuning. 2nd ed. Research Triangle Park, NC: International Society for Measurement and Control, 1995: 59–119.
6. Meng X-L, Rosenthal R, Rubin DB. Comparing correlated correlation coefficients. Psychol Bull 1992; 111: 172–5.
7. Morley AP, Chung DC, Wong AS, Short TG. The sedative and electroencephalographic effects of regional anaesthesia. Anaesthesia 2000; 55: 864–9.
8. Tverskoy M, Shagal M, Finger J, Kissin I. Subarachnoid bupivacaine blockade decreases midazolam and thiopental hypnotic requirements. J Clin Anesth 1994; 6: 487–90.
9. Tverskoy M, Fleyshman G, Bachrak L, Ben-Shlomo I. Effect of bupivacaine-induced spinal block on the hypnotic requirement of propofol. Anaesthesia 1996; 51: 652–3.
10. Kiyama S, Takeda J. Effect of extradural analgesia on the paradoxical arousal response of the electroencephalogram. Br J Anaesth 1997; 79: 750–3.
11. Kiyama S, Tsuzaki K. Processed electroencephalogram during combined extradural and general anaesthesia. Br J Anaesth 1997; 78: 751–3.
12. Ghouri AF, Monk TG, White PF. Electroencephalogram spectral edge frequency, lower esophageal contractility, and autonomic responsiveness during general anesthesia. J Clin Monit 1993; 9: 176–85.
13. White PF, Boyle WA. Relationship between hemodynamic and electroencephalographic changes during general anesthesia. Anesth Analg 1989; 68: 177–81.
14. Vernon JM, Lang E, Sebel PS, Manberg P. Prediction of movement using bispectral electroencephalographic analysis during propofol/alfentanil or isoflurane/alfentanil anesthesia. Anesth Analg 1995; 80: 780–5.
15. Leslie K, Sessler DI, Schroeder M, Walters K. Propofol blood concentration and the bispectral index predict suppression of learning during propofol/epidural anesthesia in volunteers. Anesth Analg 1995; 81: 1269–74.
16. Liu J, Singh H, White PF. Electroencephalogram bispectral analysis predicts the depth of midazolam-induced sedation. Anesthesiology 1996; 84: 64–9.
17. Liu J, Singh H, White PF. Electroencephalographic bispectral index correlates with intraoperative recall and depth of propofol-induced sedation. Anesth Analg 1997; 84: 185–9.
18. Katoh T, Suzuki A, Ikeda K. Electroencephalographic derivatives as a tool for predicting the depth of sedation and anesthesia induced by sevoflurane. Anesthesiology 1998; 88: 642–50.
19. Hans P, Lecoq JP, Brichant JF, et al. Effect of epidural bupivacaine on the relationship between the bispectral index and end-expiratory concentrations of desflurane. Anaesthesia 1999; 54: 899–908.
20. Glass PS, Bloom M, Kearse L, et al. Bispectral analysis measures sedation and memory effects of propofol, midazolam, isoflurane and alfentanil in healthy volunteers. Anesthesiology 1997; 86: 836–47.
21. Myles PS, McLeod ADM, Hunt JO, Fletcher H. Sex differences in speed of emergence and quality of recovery after anaesthesia: cohort study. BMJ 2001; 322: 710–1.
22. Schwilden H, Stoeckel H, Schüttler J. Closed-loop feedback control of methohexital anesthesia by quantitative EEG analysis in humans. Anesthesiology 1987; 67: 341–7.
23. Schwilden H, Schüttler J, Stoeckel H. Closed-loop feedback control of propofol anaesthesia by quantitative EEG analysis in humans. Br J Anaesth 1989; 62: 290–6.
© 2002 International Anesthesia Research Society
24. Mortier E, Struys M, De Smet T, et al. Closed-loop administration of propofol using bispectral analysis. Anaesthesia 1998; 53: 749–54.