1 Medical Physics and Bioengineering and 2 University Department of Anaesthesia, University of Bristol, Bristol, UK. 3 Department of Anaesthesia, Frenchay Hospital, Bristol, UK
*Corresponding author: Department of Medical Physics, Bristol General Hospital, Bristol BS1 6SY, UKE-mail: mark.a.tooley@bris.ac.uk
Accepted for publication: August 22, 2003
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods. We studied the MLAER in 27 women. They received an infusion of alfentanil 15 µg kg1 h1, followed by either a high or low infusion regimen of propofol (final infusion rates 6 and 3 mg kg1 h1). We compared the results with those of our study using propofol alone. We collected the data from two electrode sites: vertexinion and vertexmastoid. We evaluated the occurrence of the PAR and the shape of the MLAER at each electrode site.
Results. The infusion rate of propofol associated with loss of the eyelash response in 50% of subjects was 3.3 mg kg1 h1. This was significantly lower than using propofol alone (5.8 mg kg1 h1). Nb latency was the best MLAER discriminator of unconsciousness (sensitivity 94%, specificity 88%), with a threshold of 46 ms (propofol alone was 53 ms). The addition of alfentanil did not alter the relationship between propofol infusion rate and MLAER. The vertexinion electrode site gave the best protection against PAR in awake subjects (P=0.0003), and after 30 min of propofol infusion (P=0.06). The magnitude of the MLAER obtained from the vertexmastoid electrodes was larger than from the other site, although the increase was not consistent throughout the waveform (brain stem 100%, Nb 14%).
Conclusions. Addition of alfentanil lowers the propofol infusion rate required to produce unconsciousness and the Nb latency that predicts it. The better of the two sites to reduce the incidence of PAR is the vertexinion electrode site.
Br J Anaesth 2004; 92: 2532
Keywords: anaesthesia, depth; anaesthesia i.v., propofol; anaesthetic techniques, i.v.; analgesics opioid, alfentanil; monitoring, auditory evoked potentials; monitoring, auditory evoked responses; monitoring, post-auricular response
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We have examined previously the doseresponse relationship of propofol as a single agent on the MLAER, in the absence of surgical stimulation.1 In this paper, we investigate the effect of the addition of alfentanil in a similar controlled manner. We selected two infusion schemes of propofol that we predicted from the previous study would result in two distinct levels of sedation or anaesthesiaideally one leaving the patient awake and the other the patient anaesthetized. We also found from this previous work that at least 20 min of propofol infusion was required before stable anaesthesia occurred. To obtain reliable responses, the MLAER was therefore recorded at 35 min. Myogenic responses to the auditory stimulus, especially the post-auricular response (PAR),6 can be many times larger than the MLAER and can obscure it, making analysis impossible. We therefore wanted to compare the MLAER obtained from two different electrode sites and see if this affected either the shape of the MLAER or the prevalence of the PAR.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We attached two sets of bipolar silver silverchloride electrodes to the patients heads in a conventional configuration.7 One pair was placed on the vertex (positive) and 2 cm above the inion (occipital bone at the base of the skull); the other was placed on the vertex and the mastoid. A forehead electrode was used as a reference. The impedance was ensured to be below 5 k. We applied the electrodes on the ward before bringing the patient to the operating theatre.
In the anaesthetic room, we sited an i.v. cannula and began monitoring ECG, automatic non-invasive arterial pressures and pulse oximetry. We recorded baseline MLAER when each patient was awake, relaxed, and with their eyes closed.
At the beginning of the study (time zero), we gave each patient a loading dose of alfentanil 15 µg kg1, followed by an infusion of 15 µg kg1 h1. At 5 min, we started one of the two propofol infusion schemes in alternate patients. Group H (high-dose group) received a loading dose of 1 mg kg1 (given at 600 ml h1) followed by a stepped infusion of 10 mg kg1 h1 for 10 min, 8 mg kg1 h1 for 10 min, and a final rate of 6 mg kg1 h1. Group L (low-dose group) received exactly half the loading dose and infusion rates. Automatic non-invasive arterial pressure, heart rate, eyelash reflex, and response to the command squeeze my hand were recorded at 5-min intervals from time zero. MLAER was recorded at 35 min. Immediately afterwards, withdrawal response to venepuncture was noted when blood was taken for propofol assay.
The blood was collected in oxalate tubes and stored at 4°C. Propofol assay was carried out using high performance liquid chromatography.8 At the end of the study, the patient was taken into the operating theatre and anaesthesia and surgery continued as planned.
MLAER processing
We used similar equipment to that in our previous study.1 A personal computer was used to control two signal processing microprocessors (CED 1401 plus, Cambridge Electronic Design Ltd, Cambridge), one as a programmable auditory stimulator and the other for data acquisition. The auditory stimuli (rarefaction clicks of 0.5 ms duration) were delivered via insert headphones at 80 dB above the normal hearing threshold. Each set of recordings consisted of at least 500 sweeps of electroencephalograph (EEG). Each sweep of EEG data lasted 128 ms with the auditory stimulus occurring 28 ms after the beginning of the sweep. This would divide the sweep into two: a pre-stimulus interval of 28 ms (where there should be no response), then a MLAER in the following 100 ms. The signals were recorded using two high quality patient-isolated EEG amplifiers (CED 1902). The amplifiers had analogue second order high pass and low pass filters set to 3 dB points of 15 and 200 Hz, respectively. Data acquisition software (SIGAVG, Cambridge Electronic Design) controlled the acquisition parameters: repetition rate of the auditory stimulus (5.9 Hz), the duration of the sweep, and the sampling frequency of the EEG signals (1 kHz). A monitor displayed the real-time coherent average MLAER, and the sweeps were stored on disk for off-line processing.
The off-line processing consisted of coherent averaging. Before each sweep was added to the average, it had an artefact rejection algorithm applied. If a sweep was deemed to have gross artefacts (i.e. a non-physiological signal or 50 Hz) or not be representative of the background EEG, then it was not included. The artefact removal was done semi-automatically using custom designed signal processing software1 (SPIKE 2 language, Cambridge Electronic Design). Additionally, all sweeps were digitally filtered by a 15 Hz high pass filter to remove low frequency EEG. To gain confidence in the final average and of the filtering and artefact removal process, additional recording of 500 sweeps were made with no auditory stimulus present. It was expected that the average of the background EEG would tend to zero. Also, the pre-stimulus waveform should tend to zero. The final satisfactory estimates of the MLAER were stored as separate files for feature extraction and analysis.
Feature extraction from the MLAER
The simple features that were extracted consisted of the amplitudes (measured from the preceding peak or trough) and latencies (from the stimulus) to the peaks or troughs of the conventionally labelled components of the MLAER (Na, Pa, Nb). Also, the amplitude of the brain stem wave V (BS) was obtained for analysis. These were all obtained manually by applying cursors to the screen display of each satisfactory estimate of the MLAER.
Two empirically devised composite measures of the complete MLAER wave were also used. One was the best composite measure found in our previous study,1 which is the integral of the rectified smoothed differential of the wave form between 20 and 70 ms (differential index). The other measure used was the AER index,9 which is the rectified smoothed squared differential of the complete AER wave.
Statistical analysis
We used Students t-tests to compare ages, weights, and baseline arterial pressure in Groups L and H. We used one-way analysis of variance to examine the effect of the anaesthetic regimen on arterial pressure with time.
We examined the incidence of PAR on the MLAER and BS waveforms and compared the effect of electrode placement and propofol infusions. The PAR is a large and well-defined response, considerably larger than the MLAER, typically lasting from 10 ms until 30 ms, with a well-defined positive peak around 12 ms and a well-defined negative peak around 17 ms.6 The proportions were analysed using 2 analysis. We also investigated the effect of electrode placement of the MLAER waveform by comparing the amplitude of the simple AER measures (BS, Pa, and Nb). Significance analysis was carried out by t-tests.
We examined the effect the two propofol infusion groups (L and H) had on the derived variables of the MLAER waveform and compared the differences by t-tests.
We examined the relationship of serum propofol concentration with final propofol infusion rate using regression analysis.
Relationships between infusion rate and blood concentration, and between the value of features extracted from the MLAER, were determined by non-linear curve fitting programmes (Sigma-Plot for Windows version 8, SPSS Inc.). Logistic regression (Stata version 6) analysis was used to determine MLAER descriptors of the patients response to eyelash and venepuncture responses. Probit analysis (Stata version 6, and Arcus Quickstat) was used to estimate the 50 and 95% probabilities of no response.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Sampling and editing of the MLAER
Results from four patients were not included in the study because their MLAER data were too contaminated with 50 Hz and diathermy interference. The non-stimuli averages also did not tend to zero. Eleven patients remained in Group L and 11 patients in Group H, all of whom had satisfactory signals and all their non-stimuli averages tended to zero (including the pre-stimuli). One patient was maintained on an infusion of alfentanil. Satisfactory estimates of the awake MLAER were obtained from only 13 patients, but MLAER estimates could be obtained from all 22 patients during the propofol infusions. The final estimates required a variable degree of editing to exclude artefacts from the final coherent average.
Doseconcentration effects
The relationship between final propofol infusion rate (I) and blood propofol concentration (PC), was described by the regression equation:
PC = 0.68 * I 0.48, R2=0.55
This is shown in Figure 1. The grey regression line represents previous data and is considered in the discussion.
|
|
|
|
|
|
This relationship is linear. The grey regression lines represent previous data.1
Figure 4 shows the regression of Nb latency against blood concentration (BC) of propofol. The regression equation is:
|
The grey regression lines represent previous data.1
Doseconcentration and MLAER indices on responsiveness to stimuli
Table 4 shows the derived MLAER variables and their abilities to discriminate between awake and anaesthetized patients. The best discrimination variable was Nb latency in terms of its correctness, sensitivity, specificity, and the statistical fit. For comparison, blood concentration and final infusion rate of propofol are included in the table as discriminators. The final infusion rate provided perfect discrimination as all Group H patients were anaesthetized.
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We found that increasing the infusion rate of propofol produced measurable changes in the MLAER, with the differential index and Pa amplitude being the best discriminators. However, the best predictor of loss of consciousness was Nb latency with a threshold value of 44 ms. This is lower than in our previous study (Nb=53 ms) using propofol alone,1 but similar to Thorntons study (44.5 ms) in which a nitrous oxide background was used.10 It would appear that combining propofol with opioids or other analgesics reduces the value of Nb latency which predicts loss of consciousness. This might explain the large variation in predictive threshold values quoted in the literature for different drug combinations, or different anaesthetic drugs (which have different analgesic properties). This would suggest that MLAER might not be so useful as an absolute monitor of awareness, only as a relative one. Using propofol as an example, one could set the threshold for Nb latency to predict consciousness at the value based on data for propofol alone1 but, when the analgesic (for example, alfentanil) is added the threshold will shift to a lower value. Consequently, the value set is more likely to indicate adequate anaesthesia, with a higher margin for error. However, in this situation more drug than necessary may be used, increasing the potential for adverse clinical effects.
Our data suggest that the Nb latency (and other derivatives of the MLAER) has a relationship with the anaesthetic dose and this does not change when the analgesic (alfentanil) is added. This is shown in Figure 3, which shows the regression line (with confidence intervals) of the current data drawn with the regression line of our previous data for propofol alone, drawn in grey. The equations of these two regression lines are almost indentical:
latency (ms)=40.3+2.04 dose, R2=0.63 (propofol plus alfentanil)
latency (ms)=39.5+2.37 dose 0.03 dose2, R2=0.86 (propofol only)
Other researchers11 have agreed with the finding that the addition of alfentanil to propofol does not change the doseresponse relationship of MLAER. However, they concluded that alfentanil does not change the point at which the MLAER predicts arousal, but we have found that it does, for example the Nb latency value for loss of consciousness decreases to 46 ms. Table 4 shows the comparison of the predictive thresholds of our two studies, with and without alfentanil. This shows, for example, a reduction of the value of Nb latency that predicts arousal when alfentanil is added.
We found that less propofol was required to induce unconsciousness when alfentanil was added, and this paralleled the lower Nb latency that predicted loss of consciousness. Table 5 includes the comparison of the probability of no response with blood concentration and final infusion rate of propofol for the two studies.
Dutton and colleagues12 compared propofol and desflurane and MLAER. Their Nb latency values with propofol were similar to ours. However, their Nb latency values obtained during desflurane anaesthesia were different. This is further evidence that the Nb latency (or MLAER) values that predict arousal are agent specific.
The relationship between final propofol infusion rate and blood concentration in both our studies is interesting. Figure 1 includes the regression lines (grey lines) from our previous data of propofol only,1 and therefore demonstrates the regression of propofol blood levels related to final infusion rate for the two sets of data. Comparing the two regression lines, and their confidence limits, it would appear that when alfentanil is added to propofol, the blood concentration for a given final infusion rate is higher. The propofol assays were done by the same personnel in the same laboratory for both studies, making it unlikely that this finding is a result of error in the assay. This phenomenon has been noted before.13 It is further demonstrated in the graphs of Nb latency against propofol blood concentrations from the two studies, shown in Figure 4. The propofol only regression lines are shown in grey. Each concentration value with alfentanil gives a lower Nb latency value than with the propofol alone. This is consistent with higher propofol blood values being present with alfentanil, as each concentration value represents a lower final infusion value, and hence Nb latency. It seems possible that a pharmacokinetic interaction between propofol and alfentanil has led to increased plasma propofol concentrations.
The second aim of our study was to investigate which is the best electrode site to use for MLAER measurement. Some researchers state that vertexmastoid is the site to use,12 whilst others use the vertexinion configuration.3 There is little consensus on this matter, although it seems agreed that the vertex site gives the best amplitude response.14 We were interested if the choice of site had any effect on the occurrence of the PAR and the magnitude and shape of the MLAER. This confounding post-auricular response occurred in 60% of our subjects, but was less likely to occur in the vertexinion configuration (P<0.05).
Decision about electrode configuration should be made with the individual situation in mind. If the experimental circumstances are looking at anaesthesia and analgesia only, then the best configuration is probably the vertexinion one. This site gives the best protection against PAR, but this configuration does suffer from reduced amplitude for the beginning of the signalthe brainstem and early waves. The amplitude of the Nb and later waves seem to be affected less. If the circumstances involve the use of a neuromuscular blocking agent, then the PAR will be less of a problem (as it is a myogenic response), and the mastoid configuration could be used. The choice of electrode site becomes more critical if a variable derived from the MLAER is used for automatic control of anaesthesia,9 or if the signal is processed to give a single value signal index,9 without any reference to the underlying waveform. A myogenic response appearing, such as the PAR, could alter the index for the wrong reason. Electrode placement tends to be quicker on the mastoid site, but we have had no problems in fixing electrodes to the inion, even when some of our patients had very full heads of hair.
The processed values (for example differential index) we have looked at previously and in this study are not as consistently predictive as the Nb latency. However, the Nb latency is normally obtained by visual inspection and it is very difficult to extract from a waveform automatically using a computer algorithm.15 Our differential value1 appears to be useful and indeed other researchers have proposed this approach.16 More recently, Kenny and colleagues9 proposed an altered version of this, using the square of the differential and using the full width of the signal. As this AER index does not filter out background EEG or brain stem AER, its lesser sensitivity and specificity are not surprising.
It is technically difficult to obtain the MLAER robustly. Awake, unpremedicated and sedated patients are the most challenging, as electromyographic (EMG) responses and PAR occur within the frequency pass band of interest and can be difficult to remove. Using the vertexinion configuration can reduce this problem. We found that great care was needed to obtain confident estimates of the MLAER, and we had constantly to ensure that we were looking at MLAER, and not artefacts, by checking with the unstimulated signals and the signals collected pre-stimulus. Both of these unstimulated signals should produce a flat MLAER trace.
Proposed monitors that present a single number to represent the clinical state are attractive, but if this number is not accompanied by a measure of confidence, or an indicator showing the waveform shape, then it may be spurious. For example, if frontal electrodes are used, and the patient becomes more aroused, then frontal EMG can appear in the unprocessed signal. The averaging process may or may not translate these signals into high frequency components in the MLAER, but this process is likely to be unpredictable and inconsistent. The result of this would be an inaccurate, inappropriately high number. MLAER processing using coherent averaging requires a stable state of anaesthesia. Therefore, the MLAER is less useful when there are rapidly changing drug levels. There are techniques becoming available using single MLAER estimates, however, which are promising.16 Our work suggests that a monitor would have to have some knowledge of which drug or drug combination was being used and a reference table used to provide different thresholds for prediction. If the thresholds were to be based on the anaesthetic drug only studies, then this would provide maximum sensitivity, but lower specificity.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2 Thornton C, Konieczko KM, Knight AB, et al. Effect of propofol on the auditory evoked response and oesophageal contractility. Br J Anaesth 1989; 63: 41117[Abstract]
3 Newton DE, Thornton C, Konieczko KM, et al. Auditory evoked response and awareness: a study in volunteers at sub-MAC concentrations of isoflurane. Br J Anaesth 1992; 69: 1229[Abstract]
4 Schwender D, Golling W, Klasing S, et al. Effects of surgical stimulation on midlatency auditory evoked potentials during general anaesthesia with propofol/fentanyl, isoflurane/fentanyl and flunitrazepam/fentanyl. Anaesthesia 1994; 49: 5728[ISI][Medline]
5 Thornton C. Evoked potentials in anaesthesia. Eur J Anaesthesiol 1991; 8: 89107[ISI][Medline]
6 Picton TW, Hillyard SA, Krause HI, et al. Human auditory evoked potentials. I: evaluation of components. Electroen Clin Neuro 1974; 36: 17990[CrossRef][ISI]
7 Regan D. Electrodes and amplifiers. In: Regan D, ed. Human Brain Electrophysiology. New York: Elsevier, 1986; 821
8 Plummer GF. Improved method for the determination of propofol in blood by high-performance liquid chromatography with fluorescence detection. J Chromatogr 1987; 421: 1716[Medline]
9 Kenny GN, Mantzaridis H. Closed-loop control of propofol anaesthesia. Br J Anaesth 1999; 83: 2238
10 Thornton C, Barrowcliffe MP, Konieczko KM, et al. The auditory evoked response as an indicator of awareness. Br J Anaesth 1989; 63: 1135[Abstract]
11 Iselin-Chaves IA, El Moalem HE, Gan TJ, et al. Changes in the auditory evoked potentials and the bispectral index following propofol or propofol and alfentanil. Anesthesiology 2000; 92: 130010[ISI][Medline]
12 Dutton RC, Smith WD, Rampil IJ, et al. Forty-hertz midlatency auditory evoked potential activity predicts wakeful response during desflurane and propofol anesthesia in volunteers. Anesthesiology 1999; 91: 120920[ISI][Medline]
13 Pavlin DJ, Coda B, Shen DD, et al. Effects of combining propofol and alfentanil on ventilation, analgesia, sedation, and emesis in human volunteers. Anesthesiology 1996; 84: 2337[ISI][Medline]
14 Starr A, Don M. In: Picton TW, ed. Human Event-Related Potentials. Amsterdam: Elsevier, 1988; 97158
15 Gemal A. Changes in the auditory middle latency response to propofol infusion. Bristol University; 1999
16 Jensen EW, Litvan H. Detection of level of consciousness during propofol anaesthesia by rapidly extracted auditory evoked potentials. In: Vaughan DJA, Jordan C, Newton DEF, eds. Memory and Awareness in Anaesthesia 4. London: ICP, 2000; 8896